CN113395415A - Camera data processing method and system based on noise reduction technology - Google Patents
Camera data processing method and system based on noise reduction technology Download PDFInfo
- Publication number
- CN113395415A CN113395415A CN202110942256.4A CN202110942256A CN113395415A CN 113395415 A CN113395415 A CN 113395415A CN 202110942256 A CN202110942256 A CN 202110942256A CN 113395415 A CN113395415 A CN 113395415A
- Authority
- CN
- China
- Prior art keywords
- image
- data
- unit
- window
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
- H04N23/81—Camera processing pipelines; Components thereof for suppressing or minimising disturbance in the image signal generation
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
- Facsimile Image Signal Circuits (AREA)
- Picture Signal Circuits (AREA)
Abstract
The invention discloses a camera data processing method and system based on a noise reduction technology, and aims to solve the technical problems that image noise cannot be processed in the prior art, the noise can influence the whole process of image processing, acquisition and output, the definition of a picture is reduced, the quality of the picture is low, data is not stored in a blocking mode, and the utilization rate of a storage space is reduced. The camera data processing method and system based on the noise reduction technology comprises the following steps: extracting signals, converting the signals into digital images, denoising, image segmentation, enhancing the usefulness of the images, extracting moving objects, partitioning data and storing the partitioned data. The camera data processing method and system based on the noise reduction technology can perform complex nonlinear processing, solve the problem that the image quality of an actual image is reduced due to various parasitic effects, and effectively improve the image quality through the data processing module, so that the definition and quality of the image are improved, and the utilization rate of a storage space is improved.
Description
Technical Field
The invention belongs to the technical field of camera data processing, and particularly relates to a camera data processing method and system based on a noise reduction technology.
Background
The camera integrates components of image information conversion, storage, transmission and the like, has a digital access mode, is convenient to interact with a computer for processing, and can be widely used in a plurality of fields due to the continuous development of an electronic imaging technology at any time.
At present, the invention patent with patent number CN112437224A discloses a camera data processing method, which includes: capturing current frame camera image data called back by a camera, wherein the current frame camera image data is a frame of image data currently acquired by the camera; decoding a specified data source, and acquiring one frame of specified image data in the specified data source; replacing the current frame image data with the one frame designated image data; and performing operation corresponding to the current frame image data by using the specified image data.
In yet another aspect, the present application provides a camera data processing system, comprising: the data acquisition module is used for acquiring current frame camera image data called back by a camera, wherein the current frame camera image data is a frame of image data currently acquired by the camera; the data source decoding module is used for decoding a specified data source and acquiring one frame of specified image data in the specified data source; the data replacement module is used for replacing the current frame image data with the frame designated image data; and the data processing module is used for carrying out operation corresponding to the current frame image data by utilizing the specified image data. In yet another aspect, the present application further provides a computer device, including: a processor and a memory; wherein the processor is configured to execute a program stored in the memory; the memory is to store a program to at least: capturing current frame camera image data called back by a camera, wherein the current frame camera image data is a frame of image data currently acquired by the camera; decoding a specified data source, and acquiring one frame of specified image data in the specified data source; replacing the current frame image data with the one frame designated image data; and performing operation corresponding to the current frame image data by using the specified image data. The data processing method does not make any requirement on data actually acquired by a camera, avoids dependence on an actual scene when image data is acquired, cannot process image noise, the noise can influence the whole process of image processing, acquisition and output, the definition of a picture is reduced, the quality of the picture is low, the data is not stored in a blocking mode, and the utilization rate of a storage space is reduced.
Therefore, it is necessary to solve the problem that the camera data does not have the noise reduction function, so as to improve the use scene of the camera.
Disclosure of Invention
(1) Technical problem to be solved
The invention aims to provide a camera data processing method and system based on a noise reduction technology, aiming at solving the technical problems that image noise cannot be processed in the prior art, the noise can influence the whole process of image processing, acquisition and output, the definition of a picture is reduced, the quality of the picture is low, data is not stored in a blocking mode, and the utilization rate of a storage space is reduced.
(2) Technical scheme
(3) In order to solve the above technical problem, the present invention provides a camera data processing method based on noise reduction technology, comprising the following steps:
the method comprises the following steps: extracting corresponding initial image data signals, wherein the initial images are analog images, the signal comparison unit compares the analog signals with different reference voltages Ui for a plurality of times to enable the converted digital quantity to gradually approach to the corresponding value of the analog signals in numerical value, firstly, the highest bit of the register unit is set to be 1 by the time sequence control unit to enable the output digit to be 100 … … 0, the output digit is converted into the corresponding analog voltage U0 to be compared with the Ui, and if the U0 is more than the Ui, the highest bit 1 is cleared; if U0 < Ui, keeping the highest 1, recording as I, counting by the counting unit, repeating I = I +1, until the lowest bit, the state of the counting unit is the required digital quantity output, and obtaining the digital image;
step two: sending the converted digital image to a data processing module, carrying out denoising processing on the digital image, and assuming that the output and the input of a guide filter function meet a linear relation in a two-dimensional window based on a guide filter algorithm, as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is obvious that a =1 and b =0 is the solution with E (a, b) as the minimum value, the filter has no effect at this time, and the input is the output which is not fixed; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiWhen it is necessary to include all the pointsLinear function value is averaged, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we will obtain two images a in the previous stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain an output image q after final filtering;
step three: the digital image is divided into a plurality of mutually disjoint areas by a data processing module, so that the areas have consistency, the attribute characteristics of the adjacent areas have obvious difference, and the characteristics or the areas in the image are extracted;
step four: restoring the image to the original image under visual perception through an image data enhancement unit, enhancing the required information in the image and inhibiting other unnecessary information;
step five: detecting a change area through an image data target detection and motion detection unit, and extracting a motion target from a background image;
step six: the block unit in the data storage module divides the file to be stored into data blocks with variable length, the length of the data block is between a specified minimum value and a specified maximum value, the data blocks with variable length are divided by a sliding window, a sub-data block K1 is created when the Hash value of the sliding window is matched with a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is f at a position K, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked to obtain sub-data blocks K2 and K3 … … Kn, each of the divided blocks calculates its fingerprint value by using a Hash function to compare with the stored sub-data blocks, if the same fingerprint value is detected, deleting the sub data block represented by the sub data block, otherwise, storing a new sub data block;
step seven: a plurality of storage spaces S1, S2 and S3 … … Sn are divided in storage units in the data storage module, a sub data block K1 in a block unit is compared with the size of the residual space in S1, K1 is stored in S1 when K1 is smaller than the residual space in S1, K1 is continuously compared with the size of the residual space in S2 when K1 is larger than the residual space in S1, K1 is stored in S2 when K1 is smaller than the residual space in S2, K1 is continuously compared with the size of the residual space in S3 when K1 is larger than the residual space in S2 until the storage of K1 is completed, and K2 and K3 … … Kn are sequentially stored in S1, S2 and S3 … … Sn.
A camera data processing system based on noise reduction technology comprises the camera data processing method based on noise reduction technology, and the system comprises a data conversion module, a data processing module and a data storage module, wherein the data conversion module comprises a ramp signal generation unit, a signal comparison unit, a counting unit, a time sequence control unit and a registering unit, the ramp signal generation unit converts the level of a sampling signal into a ramp signal of time axis length information, the signal comparison unit compares the ramp with the sampling signal, the counting unit generates global title codes, the global title codes are counted in an increasing mode along with the reduction of the ramp signal, the time sequence control unit is used for controlling the time points of ramp signal generation and global title code generation, and the data processing module comprises an image data denoising unit, an image segmentation unit, an image data enhancement unit and an image data target detection and motion detection unit, the image data denoising unit is used for solving the problem of image quality reduction of actual images due to various parasitic effects, the image quality can be effectively improved and the signal-to-noise ratio can be increased through the image data denoising unit, the image segmentation unit divides the image into a plurality of mutually disjoint regions to ensure that the regions have consistency, and the attribute features between adjacent regions have obvious difference, the image data enhancement unit can improve the fuzzy condition of the image, emphasize and enhance the locality of the image, has better visual effect, the image data object detection and motion detection unit is adapted to identify points in the digital image where the brightness variation is significant, the data storage module comprises a blocking unit and a storage unit, wherein the blocking unit is used for blocking the image data, and the storage unit stores the data blocks blocked by the blocking unit.
Preferably, the image segmentation unit adopts an image segmentation method combining an edge and a region, avoids over-segmentation of the region by limitation of edge points, and supplements a missing edge by region segmentation to make a contour more complete, wherein the graph segmentation feature may be an original feature of the image, such as a gray value of a pixel, an object contour, a color, a reflection feature, a texture, and the like, or a spatial spectrum, such as a histogram feature.
Preferably, the detection method preset in the image data target detection and motion detection unit is a background subtraction technique, which uses the difference between the current image and the background image to detect the motion region, the image data enhancement unit includes spatial domain processing and frequency domain processing, the spatial domain processing method is to directly process the pixels in the image, based on gray scale mapping transformation, enhance the contrast of the image, improve the internal hierarchy of the image, during the frequency domain processing, fourier transform is applied, the image frequency spectrum is modified by a digital filtering method, and then inverse transformation is performed to obtain the enhanced image.
Preferably, a reference voltage Ui is preset in the signal comparison unit, the signal comparison unit compares the analog signal with different reference voltages Ui for multiple times, so that the converted digital quantity is gradually close to the corresponding value of the analog signal in value, firstly, the highest bit of the register unit is set to 1 by the timing control unit, so that the output digit is 100 … … 0, the output digit is converted into a corresponding analog voltage U0, the comparison is performed with Ui, and if U0 > Ui, the description number is too large, the highest bit 1 is cleared; if U0 < Ui, the highest bit 1 is retained, denoted as I, the counting unit counts, and so on, I = I +1, until the lowest bit, the state of the counting unit is the required digital output.
Preferably, the image data denoising unit obtains a new central pixel value by analyzing direct relation between a central pixel and other adjacent pixels in a gray scale space within a window of a certain size, and a guided filtering algorithm is preset in the image data denoising unit, and it is assumed that output and input of the guided filtering function satisfy a linear relationship within a two-dimensional window, as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is obvious that a =1 and b =0 is the solution with E (a, b) as the minimum value, the filter has no effect at this time, and the input is the output which is not fixed; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiThen, it is only necessary to average all the linear function values containing the point, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we will obtain two images a in the previous stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain the final filtered output image q.
Preferably, a blocking rule is preset in the blocking unit, the file to be stored is divided into data blocks with variable length, the length of the data block is between a specified minimum value and a specified maximum value, the data blocks with variable length are divided by a sliding window, a sub-data block K1 is created when the Hash value of the sliding window matches a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is f at position K, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked to obtain sub-data blocks K2 and K3 … … Kn, each of the divided blocks calculates its fingerprint value by using a Hash function to compare with the stored sub-data block, if the same fingerprint value is detected, the sub data block it represents is deleted, otherwise the new sub data block is stored.
Preferably, the storage unit is divided into a plurality of storage spaces S1, S2, S3 … … Sn, the sub data block K1 in the block unit is compared with the size of the remaining space in S1, when K1 is smaller than the remaining space in S1, K1 is stored in S1, when K1 is larger than the remaining space in S1, K1 is continuously compared with the size of the remaining space in S2, when K1 is smaller than the remaining space in S2, K1 is stored in S2, when K1 is larger than the remaining space in S2, K1 is continuously compared with the size of the remaining space in S3 until the storage of K1 is completed, and K2, K3 … … Kn are stored in S1, S2, S3 … … Sn in sequence.
(3) Advantageous effects
Compared with the prior art, the invention has the beneficial effects that: the camera data processing method and system based on the noise reduction technology convert the analog image into the digital image by using the data conversion module, have high processing precision and rich processing content, can perform complex nonlinear processing, have flexible flexibility and are convenient for processing the image, the noise reduction processing is performed on the digital image by using the data processing module, the problem of image quality reduction of the actual image due to various parasitic effects is solved, the quality of the image can be effectively improved by using the data processing module, the signal-to-noise ratio is increased, the information carried by the original image is better embodied, the image file to be processed is improved, the definition and the quality degree of the image are improved, the processed image file is blocked and stored by using the data storage module, the size of the blocked image file is variable, and the utilization rate of a storage space is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the technical solutions in the prior art will be briefly described below, it is obvious that the drawings in the following description are only one embodiment of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a schematic view of a workflow of a camera data processing method based on a noise reduction technology according to the present invention;
fig. 2 is a schematic diagram of a frame structure of a camera data processing system based on a noise reduction technology according to the present invention.
Detailed Description
In order to make the technical means, the original characteristics, the achieved purposes and the effects of the invention easily understood and obvious, the technical solutions in the embodiments of the present invention are clearly and completely described below to further illustrate the invention, and obviously, the described embodiments are only a part of the embodiments of the present invention, but not all the embodiments.
Example 1
The present embodiment is a camera data processing method and system based on noise reduction technology, the work flow diagram of the method is shown in fig. 1, and the steps are as follows:
the method comprises the following steps: extracting corresponding initial image data signals, wherein the initial images are analog images, the signal comparison unit compares the analog signals with different reference voltages Ui for a plurality of times to enable the converted digital quantity to gradually approach to the corresponding value of the analog signals in numerical value, firstly, the highest bit of the register unit is set to be 1 by the time sequence control unit to enable the output digit to be 100 … … 0, the output digit is converted into the corresponding analog voltage U0 to be compared with the Ui, and if the U0 is more than the Ui, the highest bit 1 is cleared; if U0 < Ui, keeping the highest 1, recording as I, counting by the counting unit, repeating I = I +1, until the lowest bit, the state of the counting unit is the required digital quantity output, and obtaining the digital image;
step two: sending the converted digital image to a data processing module, carrying out denoising processing on the digital image, and assuming that the output and the input of a guide filter function meet a linear relation in a two-dimensional window based on a guide filter algorithm, as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is obvious that a =1 and b =0 is the solution with E (a, b) as the minimum value, the filter has no effect at this time, and the input is the output which is not fixed; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiThen, it is only necessary to average all the linear function values containing the point, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we will obtain two images a in the previous stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain an output image q after final filtering;
step three: the digital image is divided into a plurality of mutually disjoint areas by a data processing module, so that the areas have consistency, the attribute characteristics of the adjacent areas have obvious difference, and the characteristics or the areas in the image are extracted;
step four: restoring the image to the original image under visual perception through an image data enhancement unit, enhancing the required information in the image and inhibiting other unnecessary information;
step five: detecting a change area through an image data target detection and motion detection unit, and extracting a motion target from a background image;
step six: the block unit in the data storage module divides the file to be stored into data blocks with variable length, the length of the data block is between a specified minimum value and a specified maximum value, the data blocks with variable length are divided by a sliding window, a sub-data block K1 is created when the Hash value of the sliding window is matched with a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is f at a position K, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked to obtain sub-data blocks K2 and K3 … … Kn, each of the divided blocks calculates its fingerprint value by using a Hash function to compare with the stored sub-data blocks, if the same fingerprint value is detected, deleting the sub data block represented by the sub data block, otherwise, storing a new sub data block;
step seven: a plurality of storage spaces S1, S2 and S3 … … Sn are divided in storage units in the data storage module, a sub data block K1 in a block unit is compared with the size of the residual space in S1, K1 is stored in S1 when K1 is smaller than the residual space in S1, K1 is continuously compared with the size of the residual space in S2 when K1 is larger than the residual space in S1, K1 is stored in S2 when K1 is smaller than the residual space in S2, K1 is continuously compared with the size of the residual space in S3 when K1 is larger than the residual space in S2 until the storage of K1 is completed, and K2 and K3 … … Kn are sequentially stored in S1, S2 and S3 … … Sn.
A camera data processing system based on noise reduction technology comprises the camera data processing method based on noise reduction technology, and the system comprises a data conversion module, a data processing module and a data storage module, wherein the data conversion module comprises a ramp signal generation unit, a signal comparison unit, a counting unit, a time sequence control unit and a register unit, the ramp signal generation unit converts the level of a sampling signal into a ramp signal of time axis length information, the signal comparison unit compares the ramp with the sampling signal, the counting unit generates global title codes, the global title codes are counted in an increasing mode along with the reduction of the ramp signal, the time sequence control unit is used for controlling the time points of generating the ramp signal and generating the global title codes, the data processing module comprises an image data denoising unit, an image segmentation unit, an image data enhancement unit and an image data target detection and motion detection unit, the image data denoising unit is used for solving the problem that the image quality of an actual image is reduced due to various parasitic effects, the image quality can be effectively improved through the image data denoising unit, the signal to noise ratio is increased, the image is divided into a plurality of mutually disjoint areas by the image segmentation unit, the areas have consistency, the attribute characteristics between the adjacent areas have obvious difference, the image data enhancement unit can improve the fuzzy condition of the image, the locality of the image is emphasized and enhanced, the visual effect is better, the image data target detection and motion detection unit is used for identifying points with obvious brightness change in the digital image, the data storage module comprises a blocking unit and a storage unit, the blocking unit is used for blocking the image data, and the storage unit stores the data blocks blocked by the blocking unit.
The image segmentation unit adopts an image segmentation method combining edges and regions, avoids over-segmentation of the regions through limitation of edge points, supplements missed edges through region segmentation to enable the contours to be more complete, wherein the image segmentation characteristics can be original characteristics of the image, such as gray values, object contours, colors, reflection characteristics, textures and the like of pixels, and can also be space frequency spectrums and the like, such as histogram characteristics, a detection method preset in the image data target detection and motion detection unit is a background reduction technology, the motion region is detected by using a difference meter of a current image and a background image, the image data enhancement unit comprises space domain processing and frequency domain processing, the space domain processing method is to directly process the pixels in the image, the gray mapping is used as a basis to transform, the contrast of the image is enhanced, the internal level of the image is improved, and Fourier transform is used during the space domain processing, and modifying the image frequency spectrum by using a digital filtering method, and performing inverse transformation to obtain an enhanced image.
Meanwhile, a reference voltage Ui is preset in the signal comparison unit, the signal comparison unit compares the analog signal with different reference voltages Ui for a plurality of times, the digital quantity obtained by conversion is enabled to gradually approach to the corresponding value of the analog signal on the numerical value, firstly, the highest bit of the register unit is set to be 1 by the time sequence control unit, the output digit is 100 … … 0, the output digit is converted into the corresponding analog voltage U0 and compared with the Ui, if U0 is more than the Ui, the highest bit is cleared from 1; if U0 < Ui, the highest bit 1 is reservedThe method comprises the following steps that I, a counting unit counts, and so on, I = I +1, until the lowest bit, the state of the counting unit is the required digital quantity output, an image data denoising unit obtains a new central pixel value by analyzing direct connection between a central pixel and other adjacent pixels in a gray scale space in a window with a certain size, a guide filtering algorithm is preset in the image data denoising unit, and firstly, the output and the input of a guide filtering function are supposed to meet a linear relation in a two-dimensional window as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is obvious that a =1 and b =0 is the solution with E (a, b) as the minimum value, the filter has no effect at this time, and the input is the output which is not fixed; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiThen, it is only necessary to average all the linear function values containing the point, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we will obtain two images a in the previous stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain the final filtered output image q.
| Outputting digital items | Analog voltage | Reference voltage | U0＞Ui | U0＜Ui |
| 100 | U0 | Ui | Clearing the highest bit 1 | I is 1 |
| 99 | U0 | Ui | Clearing the highest bit 1 | I=I+1 |
|
|
|
|
|
|
| 1 | U0 | Ui | Clearing the highest bit 1 | I=I+1 |
| 0 | U0 | Ui | Clearing the highest bit 1 | I=I+1 |
TABLE 1
In addition, a block unit is preset with a block rule, a file to be stored is divided into data blocks with variable length, the length of the data block is between a specified minimum value and a specified maximum value, the data blocks with variable length are divided by a sliding window, a sub data block K1 is created when the Hash value of the sliding window is matched with a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is f at a position K, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked to obtain sub data blocks K2 and K3 … … Kn, each of the divided blocks calculates its fingerprint value by a Hash function to compare with the stored sub data block, if the same fingerprint value is detected, the sub data block it represents is deleted, otherwise the new sub data block is stored.
In addition, a plurality of storage spaces S1, S2 and S3 … … Sn are divided in the storage unit, a sub data block K1 in the block unit is compared with the size of the residual space in S1, K1 is stored in S1 when K1 is smaller than the residual space in S1, K1 is continuously compared with the size of the residual space in S2 when K1 is larger than the residual space in S1, K1 is stored in S2 when K1 is smaller than the residual space in S2, K1 is continuously compared with the size of the residual space in S3 when K1 is larger than the residual space in S2 until the storage of K1 is completed, and K2 and K3 … … Kn are sequentially stored in S1, S2 and S3 … … Sn in the same way.
A schematic diagram of a system framework structure of the camera data processing method and system based on the noise reduction technology is shown in fig. 2.
Having thus described the principal technical features and basic principles of the invention, and the advantages associated therewith, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Furthermore, it should be understood that although the present description is described in terms of various embodiments, not every embodiment includes only a single embodiment, and such descriptions are provided for clarity only, and those skilled in the art will recognize that the embodiments described herein can be combined as a whole to form other embodiments as would be understood by those skilled in the art.
Claims (7)
1. The camera data processing method based on the noise reduction technology is characterized by comprising the following steps: the method comprises the following steps:
the method comprises the following steps: extracting corresponding initial image data signals, wherein the initial images are analog images, the signal comparison unit compares the analog signals with different reference voltages Ui for a plurality of times to enable the converted digital quantity to gradually approach to the corresponding value of the analog signals in numerical value, firstly, the highest bit of the register unit is set to be 1 by the time sequence control unit to enable the output digit to be 100 … … 0, the output digit is converted into the corresponding analog voltage U0 to be compared with the Ui, and if the U0 is more than the Ui, the highest bit 1 is cleared; if U0 < Ui, keeping the highest 1, recording as I, counting by the counting unit, repeating I = I +1, until the lowest bit, the state of the counting unit is the required digital quantity output, and obtaining the digital image;
step two: sending the converted digital image to a data processing module, carrying out denoising processing on the digital image, and assuming that the output and the input of a guide filter function meet a linear relation in a two-dimensional window based on a guide filter algorithm, as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is clear that a =1 and b =0 is the solution with E (a, b) as the minimum, then filtering is performedThe wave filter has no effect and inputs the input into the output without any movement; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiThen, it is only necessary to average all the linear function values containing the point, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we will obtain two images a in the previous stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain an output image q after final filtering;
step three: the digital image is divided into a plurality of mutually disjoint areas by a data processing module, so that the areas have consistency, the attribute characteristics of the adjacent areas have obvious difference, and the characteristics or the areas in the image are extracted;
step four: restoring the image to the original image under visual perception through an image data enhancement unit, enhancing the required information in the image and inhibiting other unnecessary information;
step five: detecting a change area through an image data target detection and motion detection unit, and extracting a motion target from a background image;
step six: the block unit in the data storage module divides the file to be stored into data blocks with variable length, the length of the data block is between a specified minimum value and a specified maximum value, the data blocks with variable length are divided by a sliding window, a sub-data block K1 is created when the Hash value of the sliding window is matched with a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is f at a position K, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked to obtain sub-data blocks K2 and K3 … … Kn, each of the divided blocks calculates its fingerprint value by using a Hash function to compare with the stored sub-data blocks, if the same fingerprint value is detected, deleting the sub data block represented by the sub data block, otherwise, storing a new sub data block;
step seven: a plurality of storage spaces S1, S2 and S3 … … Sn are divided in storage units in the data storage module, a sub data block K1 in a block unit is compared with the size of the residual space in S1, K1 is stored in S1 when K1 is smaller than the residual space in S1, K1 is continuously compared with the size of the residual space in S2 when K1 is larger than the residual space in S1, K1 is stored in S2 when K1 is smaller than the residual space in S2, K1 is continuously compared with the size of the residual space in S3 when K1 is larger than the residual space in S2 until the storage of K1 is completed, and K2 and K3 … … Kn are sequentially stored in S1, S2 and S3 … … Sn.
2. A camera data processing system based on noise reduction technology, characterized in that it comprises the camera data processing method based on noise reduction technology as claimed in claim 1, the system comprises a data conversion module, a data processing module and a data storage module, the data conversion module comprises a ramp signal generation unit, a signal comparison unit, a counting unit, a timing control unit and a registering unit, the ramp signal generation unit converts the level of the sampling signal into a ramp signal of time axis length information, the signal comparison unit compares the ramp with the sampling signal, the counting unit generates global title code, the global title code counts up with the decrease of the ramp signal, the timing control unit is used to control the time point of ramp signal generation and global title code generation, the data processing module comprises an image data denoising unit, an image segmentation unit, a frame buffer, The image data enhancement unit can improve the blurring condition of the image and emphasize and enhance the locality of the image, the image data target detection and motion detection unit is used for identifying points with obvious brightness change in the digital image, the data storage module comprises a blocking unit and a storage unit, the blocking unit is used for blocking the image data, and the storage unit stores the data blocks blocked by the blocking unit.
3. The system of claim 2, wherein the detection method preset in the image data object detection and motion detection unit is a background subtraction technique, which uses a difference between a current image and a background image to detect a motion region, and the image data enhancement unit comprises spatial processing and frequency domain processing.
4. The camera data processing system based on noise reduction technology as claimed in claim 2, wherein the signal comparison unit is preset with a reference voltage Ui, the signal comparison unit compares the analog signal with different reference voltages Ui for a plurality of times, so that the converted digital quantity is numerically and successively approximated to the corresponding value of the analog signal, first, the timing control unit sets the highest bit of the register unit to 1, the output number is 100 … … 0, the output number is converted into the corresponding analog voltage U0, and compared with Ui, if U0 > Ui, the highest bit 1 is cleared if the output number is too large; if U0 < Ui, the highest bit 1 is retained, denoted as I, the counting unit counts, and so on, I = I +1, until the lowest bit, the state of the counting unit is the required digital output.
5. The camera data processing system based on noise reduction technology as claimed in claim 2, wherein the image data denoising unit obtains a new central pixel value by analyzing direct connection between the central pixel and other adjacent pixels in a gray scale space within a window of a certain size, and a guided filtering algorithm is preset in the image data denoising unit, and it is assumed that the output and input of the guided filtering function satisfy a linear relationship within a two-dimensional window, as follows: q. q.si=akIi+bk,
i
k,qi=pi-niWhere q is the value of the output pixel, i.e. p the image after noise or texture removal, niRepresenting noise, I is the value of the input image, I and k are the pixel indices, a and b are the coefficients of the linear function when the window center is located at k, when the guide map is the input image, the guide filtering becomes a filter operation that keeps the edges, i.e., I = p, taking the gradient on both sides of the upper representation can result in q '= aI', i.e., when the input map I has a gradient, the output q also has a similar gradient, μkAnd
representation I in a local window wkIs the mean and variance, | ω | is the number of all pixels in the window, pk denotes p in the window wkϵ is the regularization parameter, when I = p, qi=piN can be simplified to ak=
，bk=(1-ak) μkIf ϵ =0, it is obvious that a =1 and b =0 is the solution with E (a, b) as the minimum value, the filter has no effect at this time, and the input is the output which is not fixed; if ϵ>0 in the region of small pixel intensity variation, i.e. image I in window wkIs substantially fixed, in this case
<<ϵ, then has ak0 and bk≈μkI.e. a weighted mean filtering is performed, and in the high variance region, i.e. the image I is represented in the window wkThe variation is large, and at this time we have
>>ϵ, then has ak1 and bkThe value is approximately equal to 0, the filtering effect on the image is very weak, the edge is kept, under the condition that the window size is not changed, the filtering effect is more obvious along with the increase of ϵ, and the output value q of a certain point is required to be specificiThen, it is only necessary to average all the linear function values containing the point, qi=
=
iIi+
IWhere the output value q is again related to two mean values, respectively a and b, in the window w, we shall refer toTwo images a are obtained in one stepkAnd bkBoth perform box filtering to obtain two new graphs: ai 'and bi', then multiplying the guide image Ii by ai ', and adding bi' to obtain the final filtered output image q.
6. The noise reduction technique-based camera data processing system of claim 2, wherein the blocking unit is preset with a blocking rule, the file to be stored is divided into variable-length data blocks, the length of the data blocks is between a specified minimum value and a specified maximum value, the variable-length data blocks are divided by a sliding window, a sub-data block K1 is created when the Hash value of the sliding window matches a reference value, so that the size of the data block can reach a desired distribution, two integers E and P are predefined, if the Hash value of the data in the fixed window is K, the Hash value of the data in the fixed window is f, if f mod E = P, the position is a boundary of the data block, the process is repeated until the whole file is blocked, sub-data blocks K2, K3 … … Kn are obtained, each of the obtained blocks is compared with the stored sub-data blocks by calculating its fingerprint value by a Hash function, if the same fingerprint value is detected, deleting the sub data block represented by the fingerprint value, otherwise, storing a new sub data block.
7. The noise reduction technique-based camera data processing system according to claim 6, wherein the storage unit is divided into a plurality of storage spaces S1, S2, S3 … … Sn, the sub-data block K1 in the block unit is compared with the size of the remaining space in S1, when K1 is smaller than the remaining space in S1, K1 is stored in S1, when K1 is larger than the remaining space in S1, the comparison with the size of the remaining space in S2 is continued with K1, when K1 is smaller than the remaining space in S2, K1 is stored in S2, when K1 is larger than the remaining space in S2, the comparison with the size of the remaining space in S3 with K1 is continued until the storage of K1 is completed, and, K2, K3 … … Kn are sequentially stored in S1, S2, S3 … … Sn.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110942256.4A CN113395415B (en) | 2021-08-17 | 2021-08-17 | Camera data processing method and system based on noise reduction technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110942256.4A CN113395415B (en) | 2021-08-17 | 2021-08-17 | Camera data processing method and system based on noise reduction technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113395415A true CN113395415A (en) | 2021-09-14 |
| CN113395415B CN113395415B (en) | 2021-11-26 |
Family
ID=77622798
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110942256.4A Active CN113395415B (en) | 2021-08-17 | 2021-08-17 | Camera data processing method and system based on noise reduction technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113395415B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113704803A (en) * | 2021-10-27 | 2021-11-26 | 深圳市迪壹六电子有限公司 | Computer software data encryption system based on Internet of things and encryption method thereof |
| CN116721094A (en) * | 2023-08-03 | 2023-09-08 | 济南邦德激光股份有限公司 | Visual pattern recognition method and system under complex background and laser cutting machine |
| CN117539796A (en) * | 2024-01-09 | 2024-02-09 | 深圳宏芯宇电子股份有限公司 | Electronic device and buffer memory management method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013009242A (en) * | 2011-06-27 | 2013-01-10 | Nippon Telegr & Teleph Corp <Ntt> | Image encoding method, image encoder and program thereof |
| CN103581633A (en) * | 2012-07-24 | 2014-02-12 | 索尼公司 | Image processing device, image processing method, program, and imaging apparatus |
| US20160080626A1 (en) * | 2014-09-16 | 2016-03-17 | Google Technology Holdings LLC | Computational Camera Using Fusion of Image Sensors |
| CN106228515A (en) * | 2016-07-13 | 2016-12-14 | 凌云光技术集团有限责任公司 | A kind of image de-noising method and device |
| CN110969588A (en) * | 2019-12-02 | 2020-04-07 | 杨勇 | Image enhancement method and system based on nonlinear guided filtering |
| CN111885308A (en) * | 2015-09-24 | 2020-11-03 | 高通股份有限公司 | Phase detection autofocus noise reduction |
-
2021
- 2021-08-17 CN CN202110942256.4A patent/CN113395415B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013009242A (en) * | 2011-06-27 | 2013-01-10 | Nippon Telegr & Teleph Corp <Ntt> | Image encoding method, image encoder and program thereof |
| CN103581633A (en) * | 2012-07-24 | 2014-02-12 | 索尼公司 | Image processing device, image processing method, program, and imaging apparatus |
| US20160080626A1 (en) * | 2014-09-16 | 2016-03-17 | Google Technology Holdings LLC | Computational Camera Using Fusion of Image Sensors |
| CN111885308A (en) * | 2015-09-24 | 2020-11-03 | 高通股份有限公司 | Phase detection autofocus noise reduction |
| CN106228515A (en) * | 2016-07-13 | 2016-12-14 | 凌云光技术集团有限责任公司 | A kind of image de-noising method and device |
| CN110969588A (en) * | 2019-12-02 | 2020-04-07 | 杨勇 | Image enhancement method and system based on nonlinear guided filtering |
Non-Patent Citations (2)
| Title |
|---|
| H.L.H. KACEM: "Improvement of the compression JPEG 2000 quality by Denoising Filters", 《2006 2ND INTERNATIONAL CONFERENCE ON INFORMATION & COMMUNICATION TECHNOLOGIES》 * |
| 谷元宝: "基于降采样块匹配的数字视频3D降噪算法", 《长春理工大学学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113704803A (en) * | 2021-10-27 | 2021-11-26 | 深圳市迪壹六电子有限公司 | Computer software data encryption system based on Internet of things and encryption method thereof |
| CN116721094A (en) * | 2023-08-03 | 2023-09-08 | 济南邦德激光股份有限公司 | Visual pattern recognition method and system under complex background and laser cutting machine |
| CN116721094B (en) * | 2023-08-03 | 2023-12-19 | 济南邦德激光股份有限公司 | Visual pattern recognition method and system under complex background and laser cutting machine |
| CN117539796A (en) * | 2024-01-09 | 2024-02-09 | 深圳宏芯宇电子股份有限公司 | Electronic device and buffer memory management method |
| CN117539796B (en) * | 2024-01-09 | 2024-05-28 | 深圳宏芯宇电子股份有限公司 | Electronic device and buffer memory management method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113395415B (en) | 2021-11-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gao et al. | Naturalness preserved nonuniform illumination estimation for image enhancement based on retinex | |
| Pu et al. | Fractional differential mask: a fractional differential-based approach for multiscale texture enhancement | |
| CN113395415B (en) | Camera data processing method and system based on noise reduction technology | |
| CN111369457B (en) | Remote sensing image denoising method for sparse discrimination tensor robustness PCA | |
| CN111798396A (en) | Multifunctional image processing method based on wavelet transformation | |
| CN111079764B (en) | Low-illumination license plate image recognition method and device based on deep learning | |
| WO2006102014A1 (en) | Fast graph cuts: a weak shape assumption provides a fast, exact method for graph cuts segmentation | |
| Srinivas et al. | Remote sensing image segmentation using OTSU algorithm | |
| Wu et al. | Two-level wavelet-based convolutional neural network for image deblurring | |
| Qiao et al. | Layered input GradiNet for image denoising | |
| Yadav et al. | Thresholding and morphological based segmentation techniques for medical images | |
| Satti et al. | DIBS: Distance-and intensity-based separation filter for high-density impulse noise removal | |
| Shankar et al. | Object oriented fuzzy filter for noise reduction of Pgm images | |
| Li et al. | Dynamic-clustering extreme intensity prior based blind image deblurring | |
| Feng et al. | An Improved Saliency Detection Algorithm Based on Edge Boxes and Bayesian Model. | |
| Rahman et al. | Towards scanning electron microscopy image denoising: a state-of-the-art overview, benchmark, taxonomies, and future direction | |
| Chen et al. | Infrared and visible image fusion with deep wavelet-dense network | |
| CN113674144A (en) | Image processing method, terminal equipment and readable storage medium | |
| Dubey et al. | A detailed study of digital image processing | |
| CN112541913B (en) | Image local fuzzy detection and extraction method based on column rate spectral analysis and noise robustness | |
| Hu et al. | PPMID-SSL: Denoising of Pediatric Pulmonary Medical Images Based on Self-Supervised Learning | |
| Zhang | Adaptive region-based image fusion using energy evaluation model for fusion decision | |
| Weli et al. | Digital Image Noise Reduction Based on Proposed Smoothing and Sharpening Filters | |
| Huang et al. | Multi-focus image fusion combined with cnn and algebraic multi-grid method | |
| Shi et al. | Comparative Study of Digital Instrument Image Enhancement in Complex Industrial Environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |