CN116205794A

CN116205794A - Image processing device and image processing method for contrast improvement

Info

Publication number: CN116205794A
Application number: CN202111440320.5A
Authority: CN
Inventors: 官侑萱; 李宗轩; 陈世泽
Original assignee: Realtek Semiconductor Corp
Current assignee: Realtek Semiconductor Corp
Priority date: 2021-11-30
Filing date: 2021-11-30
Publication date: 2023-06-02

Abstract

The invention provides an image processing method, which comprises the following steps: receiving an input image; performing a low-frequency partial image adjustment operation according to low-frequency information of the image of the at least one pixel unit of the input image, and adjusting a regional brightness value of the image of the at least one pixel unit; performing a high-frequency partial image adjustment operation according to a high-frequency information of the image of the at least one pixel unit of the input image, and improving image details of the image of the at least one pixel unit; and generating an output image according to the input image, the low frequency partial image adjustment operation and the high frequency partial image adjustment operation.

Description

Image processing device and image processing method for contrast improvement

Technical Field

The present invention relates to an image processing mechanism, and more particularly, to an image processing apparatus and an image processing method for contrast enhancement.

Background

Generally, although the prior art can adjust brightness between regions in an image frame to generate an image that meets the preference of a user, the prior art cannot effectively improve details of the image in the region, only can adjust gaussian filtering of a mask image, and has no other parameters for elastic adjustment, and the prior art can only adjust the brightness of the region, and has quite limited improvement of details of lines of the image, such as a grid or a railing.

Disclosure of Invention

It is therefore an object of the present invention to provide an image processing apparatus and an image processing method, which solve the above-mentioned problems.

According to an embodiment of the present invention, an image processing apparatus is disclosed. The image processing device comprises a processing circuit, wherein the processing circuit is used for receiving an input image and comprises a low-frequency part processing unit and a high-frequency part processing unit. The low-frequency part processing unit is used for performing a low-frequency part image adjusting operation according to low-frequency information of the image of the at least one pixel unit of the input image, and adjusting the regional brightness value of the image of the at least one pixel unit. The high-frequency part processing unit is used for performing high-frequency part image adjustment operation according to high-frequency information of the image of the at least one pixel unit of the input image, and improving image details of the image of the at least one pixel unit. The processing circuit generates an output image according to the input image, the low frequency partial image adjustment operation, and the high frequency partial image adjustment operation.

According to an embodiment of the present invention, there is further disclosed an image processing method including: receiving an input image; using a low frequency part processing unit to perform a low frequency part image adjusting operation according to a low frequency information of the image of the at least one pixel unit of the input image, and adjusting the area brightness value of the image of the at least one pixel unit; using a high-frequency part processing unit to perform a high-frequency part image adjustment operation according to high-frequency information of the image of the at least one pixel unit of the input image, and improving image details of the image of the at least one pixel unit; and generating an output image according to the input image, the low frequency partial image adjustment operation and the high frequency partial image adjustment operation.

Drawings

FIG. 1 is a block diagram of an image processing apparatus according to a preferred embodiment of the invention.

Fig. 2 is a schematic diagram of a three-dimensional model of a two-dimensional look-up table LUT generated by the processing circuit shown in fig. 1 when selecting the curve equation and the corresponding parameter value of the algorithm used by the low frequency portion processing unit and the curve equation and the corresponding parameter value of the algorithm used by the high frequency portion processing unit according to the embodiment of the present invention.

FIG. 3 is a schematic diagram of a three-dimensional model of a two-dimensional look-up table LUT generated when the processing circuit shown in FIG. 1 selects a curve equation and a corresponding parameter value of an algorithm used by a low frequency portion processing unit and selects a curve equation and a corresponding parameter value of an algorithm used by a high frequency portion processing unit according to an embodiment of the invention, wherein the low frequency and the high frequency are respectively adjusted by different curve equations.

Fig. 4 is another perspective view of a two-dimensional look-up table LUT generated by the processing circuit shown in fig. 1 when selecting the curve equation and the corresponding parameter values of the algorithm used by the low frequency portion processing unit and determining not to perform the image adjustment of the high frequency portion processing unit according to the embodiment of the present invention.

FIG. 5 is another perspective view of a two-dimensional look-up table LUT generated by the processing circuit shown in FIG. 1 when selecting the curve equation and the corresponding parameter values of the algorithm used by the high frequency portion processing unit and determining not to perform the image adjustment of the low frequency portion processing unit according to the embodiment of the invention.

FIG. 6 is a schematic diagram of an embodiment of a corresponding inverse S-curve.

FIG. 7 is a schematic diagram of an exemplary calculation of the value of mean_new using a curve formula corresponding to a nonlinear enhancement algorithm.

Fig. 8 is a comparison of an original input image and an output image after processing an image with a low frequency portion but a high frequency portion not adjusted with special parameters in an embodiment of the present invention.

Fig. 9 is a comparative diagram of image adjustment in which the high frequency portion is not adjusted with the special parameter and image adjustment in which the high frequency portion is adjusted with the special parameter.

FIG. 10 is a schematic diagram of an embodiment implemented with a broken line by an S-shaped curve at a fixed point.

FIG. 11 is a schematic diagram of an embodiment of an S-curve.

Detailed Description

The present invention is directed to an image processing method capable of adjusting or adjusting the brightness of each region in an input image and improving the picture details of the input image at the same time, which is capable of brightening a darker region (hereinafter referred to as a dark region) and a lighter region (hereinafter referred to as a bright region) in a received input image (e.g., a frame or an image picture), and improving the picture details of the bright region and the dark region to generate and output an output image.

Referring to fig. 1, fig. 1 is a block diagram of an image processing apparatus 100 according to a preferred embodiment of the invention. The image processing apparatus 100 includes a storage unit (e.g. a memory unit or a memory) 105 and a processing circuit 110, wherein the processing circuit 110 includes a high-frequency portion processing unit 110H and a low-frequency portion processing unit 110L. In this embodiment, the image processing apparatus 100 (or the processing circuit 110) receives an input image, such as an input frame or an input frame image, and is configured to perform at least one specific image processing operation on an image of at least one pixel unit in the input image, so as to lighten a dark area, darken the bright area, and simultaneously enhance image details of the bright area and the dark area, wherein the high-frequency portion and the low-frequency portion can respectively use different curve equations and a plurality of different parameter values when performing image adjustment; it should be noted that the selection of the curve equation and the setting of the different parameter values are not limitations of the present invention.

The image processing operation of dark area dimming and bright area dimming may be performed by the low frequency portion processing unit 110L, where the low frequency portion processing unit 110L is configured to perform a low frequency portion image adjustment operation on a low frequency information of the image of the at least one pixel unit of the input image, where the low frequency information corresponds to a regional brightness value of the image of the at least one pixel unit, for example, the regional brightness value is an average brightness value of the image of the at least one pixel unit and images of surrounding pixel units.

The image processing operation for enhancing the image frame details is performed by the high-frequency part processing unit 110H, and the high-frequency part processing unit 110H is configured to perform a high-frequency part image adjustment operation on a high-frequency information of the image of the at least one pixel unit of the input image, where the high-frequency information corresponds to a pixel value of the image of the at least one pixel unit divided by the regional brightness value.

In other words, the processing circuit 110 of the present embodiment employs the high-frequency portion processing unit 110H to perform an image processing of a high-frequency portion on the image of the at least one pixel unit to increase the detail level of the image of the pixel unit, and employs the low-frequency portion processing unit 110L to perform an image processing of a low-frequency portion on the image of the at least one pixel unit and the surrounding image to correspondingly lighten a darker image area (i.e. a dark area) in the input image and simultaneously lighten a brighter image area (i.e. a bright area) in the input image, so as to make the overall image brightness level of the input image uniform, i.e. adjust the average brightness of each area. It should be noted that in another embodiment, the adjustment of the image may be performed by using only the low-frequency portion processing unit 110L and not using the high-frequency portion processing unit 110H, or the adjustment of the image may be performed by using only the high-frequency portion processing unit 110H and not using the low-frequency portion processing unit 110L.

In this embodiment, for example, in the image processing of one pixel unit, the high-frequency portion processing unit 110H and the low-frequency portion processing unit 110L are equivalent to perform the corresponding image processing operation on the image of the pixel unit separately and independently, and then combine the results of the image processing to obtain the final image of the pixel unit, so as to generate the final pixel value of the pixel unit in the output image (for example, an output frame or an output picture image). The same applies to the image processing of other pixel units or all pixel units.

The high-frequency part processing unit 110H and the low-frequency part processing unit 110L may be implemented as a pure hardware circuit, a pure software program, or a combination of hardware and software, which is not a limitation of the present disclosure. For example, the high-frequency processing unit 110H and the low-frequency processing unit 110L may be implemented by software programs of different algorithms, and the user may determine the curve equation and the corresponding parameter value of each algorithm to determine the overall brightness uniformity and the picture detail of the final output image.

In this embodiment, for a specific pixel unit (e.g., a current pixel unit) in the input image, the image processing of the high-frequency portion may be to adjust and process the reflectivity (or brightness reflectivity) of the pixel image of the specific pixel unit, and the image processing of the low-frequency portion may be to adjust and process the brightness (or average brightness) of the pixel image of the specific pixel unit. For example, the information of the image (or pixel value) of the specific pixel unit can be regarded as a product of its low frequency information and its high frequency information:

where val_org is the image information of the specific pixel unit, for example, a pixel value, which may be, for example, in the range of 0 to 255, or may take a normalized (normalized) representation between 0 and 1, where a value of 0 may refer to a minimum pixel value (or absolute pixel value of 0) in the input image and a value of 1 may refer to a maximum pixel value (or absolute pixel value of 255) in the input image. mean_org is low frequency information of the image of the specific pixel unit, such as the area average brightness, and its value may be in the range of 0 to 255, or a normalized representation between 0 and 1 may be adopted, where the value 0 may refer to a minimum brightness value in the input image and the value 1 may refer to a maximum brightness value in the input image.

The high frequency information of the image, which is the specific pixel unit, is, for example, the reflectivity of the image, which is not necessarily 0-1, but may be greater than 1, for example, the value of val_org is greater than the value of mean_org if the current pixel is on the white side of the image, thus>

In addition, another example is an active emitter, which may have a reflectance of greater than 1.

For generating an image in the output image corresponding to the pixel unit, the image information of the pixel unit in the output image can be also split into a product of low-frequency information and high-frequency information:

wherein val_new is a resulting pixel value in the output image generated by the image processing apparatus 100 after performing image processing on the pixel value of the pixel unit in the input image; mean_new may be regarded as low frequency information of a result in the output image generated after the low frequency part processing unit 110L performs image processing of the original low frequency information mean_org of the pixel unit;

it can be regarded that the original high frequency information of the pixel unit is +.>

High frequency information of a result in the output image generated after image processing of the high frequency part. The val_new and mean_new may be normalized, which are not described herein, but +_new>

The value of (2) may be between 0 and 1, or greater than 1.

In one embodiment, after the user determines the curve equation and the corresponding parameter value of the algorithm adopted by the high frequency portion processing unit 110H and the curve equation and the corresponding parameter value of the algorithm adopted by the low frequency portion processing unit 110L, the image of one or each pixel unit in an input image can be processed accordingly to generate the image of one or each pixel unit in an output image.

In other embodiments, to simplify the calculation of the curve equation according to different parameter values, the processing circuit 110 may be configured to generate a plurality of corresponding different area average luminance values according to a plurality of different pixels, and then generate a plurality of output pixel values according to a plurality of combinations of the plurality of different pixel values and the plurality of different area average luminance values, wherein the values may be represented by normalization, for example, and generate a two-dimensional look-up table LUT according to the combinations and the generated values of the output pixels, and the two-dimensional look-up table LUT may be stored in the storage unit 105.

For example, the values of the input pixels, the area average luminance values and the generated output pixels can be normalized, respectively, with a possible value ranging from 0 to 1, and the possible range of the values of the input pixels (between 0 and 1) can be equally divided into N sections, the distance range of each section is

That is, using (n+1) segmentation points, for example, in a simple example (but not limited to), if N is equal to 4, the distance of each segment is 0.25, and the selection of 5 segmentation points may be a combination of 0, 0.25, 0.5, 0.75, 1; the possible value range (between 0 and 1) of the area average brightness value of the input pixel can be divided into M sections, and the distance range of each section is +.>

That is, (M+1) segmentation points are used, where N and M may also be the same value. For example, in an embodiment, N and M are equal to 32 (but not limited to), the processing circuit 110 may generate corresponding 33×33 output pixel values according to 33×33 combinations of 33 segmentation points, and it should be noted that some of the output pixel values may have equal or similar values, which is not a limitation of the present invention.

Therefore, when a new incoming input pixel is received, the processing circuit 110 may generate a corresponding area average luminance value according to the input pixel value and the generated area average luminance value, index and find two adjacent segmentation points in the two-dimensional lookup table, for example, find two segmentation points whose values are adjacent according to the input pixel value, find two segmentation points whose values are adjacent according to the area average luminance value, and find the values of 4 output pixels corresponding to the combination of the 4 segmentation points, and then the processing circuit 110 may perform a bilinear interpolation calculation according to the input pixel value, the generated area average luminance value, the found 4 segmentation points, and the values of the 4 output pixels to generate a final output pixel value.

Referring to fig. 2 to 5, fig. 2 to 5 are schematic three-dimensional model diagrams of the processing circuit 110 respectively showing the relationships among the (n+1) segmentation points of the input pixel value, the (m+1) segmentation points of the region average brightness value, and the (n+1) x (m+1) output pixel values by using curve equations and parameter values corresponding to different algorithms, where N and M are, for example, equal to 32. FIG. 2 is a schematic diagram of a three-dimensional model of a two-dimensional LUT generated when the processing circuit 110 shown in FIG. 1 selects a curve equation and a corresponding parameter value of an algorithm used by the low-frequency portion processing unit 110L and a curve equation and a corresponding parameter value of an algorithm used by the high-frequency portion processing unit 110H according to an embodiment of the present invention, wherein the curve equation of the algorithm used by the low-frequency portion processing unit 110L and the high-frequency portion processing unit 110H adopts a parameter transformation related to the same curve equation, and the algorithm is, for example, an algorithm of a local color correction mechanism (Local Color Correction, LCC), but is not limited thereto; it should be noted that the algorithm of the local color correction mechanism can be used to adjust the image information of the low frequency component and the high frequency component, and consistent adjustment is adopted. Fig. 3 is a schematic diagram of a three-dimensional model of a two-dimensional lookup table LUT generated when the processing circuit 110 shown in fig. 1 selects a curve equation and a corresponding parameter value of an algorithm used by the low-frequency portion processing unit 110L and selects a curve equation and a corresponding parameter value of an algorithm used by the high-frequency portion processing unit 110H according to an embodiment of the present invention, wherein the algorithm used by the low-frequency portion processing unit 110L is, for example, an algorithm of a local color correction mechanism, and the algorithm used by the high-frequency portion processing unit 110H is an algorithm of a local gamma correction (Local Gamma Correction) operation and the parameter γ is set to 1.1 (but not limited thereto). Fig. 4 is another perspective view of a two-dimensional look-up table LUT generated by the processing circuit 110 shown in fig. 1 when selecting the curve equation and the corresponding parameter values of the algorithm used by the low frequency portion processing unit 110L and determining not to perform the image adjustment of the high frequency portion processing unit 110H according to an embodiment of the present invention, wherein the algorithm used by the low frequency portion processing unit 110L is, for example, a nonlinear enhancement (Adaptive and Integrated Neighborhood-dependent Approach for Nonlinear Enhancement of Color Images, AINDANE) algorithm and the value of the parameter z is, for example, set to 0.5 (but not limited thereto). Fig. 5 is another perspective view of a two-dimensional LUT generated by the processing circuit 110 shown in fig. 1 when selecting the curve equation and the corresponding parameter values of the algorithm used by the high-frequency portion processing unit 110H and determining not to perform the image adjustment of the low-frequency portion processing unit 110L according to an embodiment of the present invention, wherein the algorithm used by the high-frequency portion processing unit 110H is, for example, an algorithm of a local gamma correction operation and the parameter γ is set to 1.5 (but not limited to).

Each intersection point of the mesh-like or lattice-like models shown in fig. 2 to 5 refers to a final output pixel value corresponding to a combination of the value of the input pixel and the corresponding region average luminance value. The models of the two-dimensional look-up tables LUTs may be stored in the storage unit 105. As can be seen from the above, in the embodiment of the present invention, the low-frequency portion processing and the high-frequency portion processing are respectively and independently executed, and then the respective results are integrated, so that the design of the image adjustment operation is more flexible, and the user can determine the algorithm curve formula and the related parameter values applicable to the high-frequency portion image and the low-frequency portion image according to the user's own needs to perform brightness adjustment and picture detail adjustment.

Regarding the operation principle of the low frequency part processing unit 110L, in one embodiment, the low frequency part processing unit 110L may use the image enhancement local contrast algorithm, such as the principle of a local color correction mechanism, to process the low frequency part of the image of a specific pixel unit, for example, taking an 8-bit RGB image (whose pixel value of R, G, B is in the range of 0 to 255) as an example, if only brightness (intensity) is considered to avoid chromatic distortion, the brightness of the image of the specific pixel unit may be an average value of the brightness of the image on R, G, B three color channels, for example:

wherein the parameter (x, y) is the pixel coordinate of the image of the specific pixel unit in the input image, I (x, y) refers to the brightness of the image of the specific pixel unit, and R (x, y), G (x, y), and B (x, y) refer to the brightness of the image of the specific pixel unit on three different color gamuts, respectively. The local color correction scheme then performs a Gaussian blur (Gaussian blur) on the image of the particular pixel cell to calculate the mask image:

M(x，y)＝(Gaussian*(255-I))(x，y)

the local color correction mechanism performs an inversion on the brightness value of the image and then performs a gaussian blur operation to obtain M (x, y), and then performs a Gamma correction (Gamma correction) on the input pixel of the specific pixel unit by using M (x, y) to generate low frequency partial image information Output (x, y) of the specific pixel unit in the Output image:

the above equation can then be reduced to a normalized version:

M′(x，y)＝(Gaussian*I)(x，y)

where Input (x, y) refers to the luminance value of the Input pixel unit at the pixel coordinate (x, y), output (x, y) refers to the luminance value of the Output pixel unit adjusted at the pixel coordinate (x, y), and M' (x, y) can be regarded as a region average luminance value of the Input pixel unit. In terms of the principle of the local color correction mechanism, when the image of the specific pixel unit is in the dark area of the image, the brightness value I (x, y) will be smaller, and the value of (255-I (x, y)) will be larger, and the value of (255-I (x, y)) after the gaussian operation will still be relatively larger, where M (x, y) has a greater chance of possibly exceeding half of the brightness value 128, the value of the gamma correction in this case will be smaller than 1, so that the correction of the value of the gamma correction smaller than 1 is performed in the dark area of the image, and the dark area of the image will be lightened accordingly because the correction curve corresponding to the value of the gamma correction smaller than 1 is an up-cast curve. In addition, when the brightness value I (x, y) is larger and the value of (255-I (x, y)) is smaller, the value of (255-I (x, y)) after the gaussian operation is still relatively smaller, and at this time, M (x, y) has a greater chance of being less than half the brightness value 128, the value of the gamma correction in this case is greater than 1, so that the gamma correction is performed at the position of the bright image, the gamma correction is greater than 1, and the bright image is accordingly dimmed accordingly because the correction curve corresponding to the gamma correction greater than 1 is a drop curve.

In addition, in order to further and more flexibly adjust the details of the image, in practice, the low frequency part processing unit 110L also employs a Fast center-to-surrounding information contrast adjustment operation (Fast center-surround Contrast Modification, FCSCM) to adjust the output image in consideration of the relationship between the image of the center-point object (pixel unit) and the images of surrounding pixel units, in addition to the above-described principle. For example, regarding the surroundings of the current pixel unit as information having an average brightness of the area, when the brightness of the surrounding image is dark, the contrast ratio of the image of the dark object and the image of the surrounding object can be pulled apart by referring to the upper-throwing curve, and when the brightness of the surrounding image is bright, the contrast ratio of the image of the bright object and the image of the surrounding object can be pulled apart by referring to the lower-throwing curve; it should be noted that the contrast adjustment operation of the fast center point and the surrounding information does not only adjust the low frequency information of the image, but also partially adjusts the high frequency information of the image. In this embodiment, taking the above normalized image as an example, the low frequency part processing unit 110L regards the luminance value Input (x, y) of the current pixel unit and the area average luminance value M' (x, y) thereof in the operation of the above local color correction mechanism as the center point information and the surrounding information thereof in the contrast adjustment operation of the fast center point and the surrounding information, respectively, so the corresponding curve equation can be expressed as follows:

where mean _ org is the area average luminance of the image for that particular pixel cell. Referring to fig. 2 again, in the model shown in fig. 2, the processing circuit 110 selects a curve equation of the local color correction algorithm used by the low frequency portion processing unit 110L, that is, consider val_org in the previous paragraph as mean_org and val_new as mean_new, so as to obtain the following equation:

this is the curve formula corresponding along the diagonal in fig. 2, i.e., the curve of fig. 6.

Referring to fig. 6, fig. 6 shows an embodiment of an inverse S-curve corresponding to the above equation, wherein the horizontal axis of the inverse S-curve is a value of mean_org, the normalized value ranges from 0 to 1, and the vertical axis is a value of mean_new, and the normalized value ranges from 0 to 1, for example, the low frequency portion processing unit 110L adjusts the brightness of the entire region in the input image according to the inverse S-curve of fig. 6, for example, when the value of mean_org is close to 0 (darkest) and close to 1 (brightest), a different parabolic curve is used to adjust the brightness contrast, and when the value of mean_org is close to 0.5 (middle brightness), the value of mean_new is quite close to the value of mean_org (the condition of mean_new=mean_org is not necessarily satisfied), so that brightness is approximately not adjusted. It should be noted that the inverse S-shaped curve is not limited in this case, and in other embodiments, curves with different shapes such as an S-shaped curve may be used, for example, referring to fig. 11, fig. 11 is a schematic diagram of an embodiment of an S-shaped curve, and implemented as a Smoothstep curve, for example, the following equation (but not limited to) may be used to implement the S-shaped curve:

mean_new＝3·mean_org ² -2·mean_org ³

the low frequency part processing unit 110L is implemented by using the inverse S-curve shown in fig. 6 to calculate the mean_new value according to the above equation using the mean_org value.

In addition, in other embodiments, the low frequency portion processing unit 110L may be implemented by using other algorithm formulas, for example (but not limited to), which may be implemented by using a curve formula corresponding to a nonlinear enhanced AINDANE algorithm to calculate the value of mean_new, for example, fig. 7 is an exemplary schematic diagram of calculating the value of mean_new using a curve formula corresponding to a nonlinear enhanced algorithm, where the nonlinear enhanced algorithm includes a parameter z, and the value of the parameter z is between 0 and 1, for example, the value of the parameter z may be set to 0.5, which is not a limitation in the present disclosure.

Furthermore, in other embodiments, the low frequency part processing unit 110L may alternatively be arranged to directly take the value of the area average luminance mean_org of all pixel units as the value of mean_new, i.e. mean_new=mean_org, without adjustment.

For image processing of the high frequency part, the high frequency part processing unit 110H adopts a local gamma correction operation and adjusts according to the following equation:

where val_org is the value of an input pixel of the specific pixel unit in the input image, γ is an adjustable parameter, and it can be seen in the logarithmic domain that γ has the effect of stretching the contrast between the pixel value of the central pixel unit and the pixel values of the surrounding pixel units, as shown in the following equation:

log(val_new)-log(mean_new)＝γ·(log(val_org)-log(mean_org))

thus, the processing device 100 recombines the adjusted low frequency and high frequency portions to:

in combination with the two-dimensional lookup table LUT, for example, 33 curves can be constructed, and γ of each curve can be adjusted, and information or values thereof can also be stored in the storage unit 105. It should be noted that the local gamma correction is not limited in this case, and in other embodiments, the high-frequency part processing unit 110H may also use other curve formulas to achieve the effect of adjusting the image details, such as an S-shaped curve.

Furthermore, in other embodiments, the high frequency portion processing unit 110H may alternatively be arranged to directly adjust the reflectivity of all pixel units without adjustment

The value of>

The value of (i), i.e

In addition, in other embodiments, the processing device 100 may also recombine the adjusted low frequency and high frequency portions into:

val_new＝min(max(m·(val_org-mean_org)+mean_new，0)，1)

where m is an adjustable slope, taking m to satisfy the following inequality when mean_org is between 0 and 1 (i.e., 0 < mean_org < 1):

and when mean_org is equal to 0 or equal to 1, taking m satisfies the following inequality:

m＞1

the equation above shows a plot of a plot, as shown in FIG. 10; FIG. 10 is a schematic diagram of an embodiment of an S-curve through a fixed point, implemented as a polyline.

Fig. 8 and 9 are provided to make the spirit of the present invention more clear to the reader. Fig. 8 is a graph comparing an original input image and an output image obtained by processing a low frequency part image and a high frequency part image according to the embodiment of the present invention by using a parametric transformation (i.e. the high frequency part adjusting operation is not additionally performed by using other parameters) of the same curve equation, as shown in the left side of fig. 8, the original input image IMG is, for example, an uneven image, for example, the whole input image IMG includes an outdoor image part OUTD and an indoor image part IND of other remaining areas, for example, the outdoor image part OUTD includes a building B1, the indoor image part IND includes an image P1 of a ladder, for example, the original input image IMG is a scene image of the daytime, which is displayed in the outdoor image part OUTD, for example, the brightness of the building B1 is too bright, the indoor image part IND includes an outdoor image part OUTD (including a window) which is a window, the outdoor image is a dark area of the most bright, the outdoor image is a dark area of the building, and the other area is a dark area of the dark area is the dark area of the building, and the light is the dark area of the dark area is the dark area. As shown on the right side of fig. 8, in an output image IMG' generated after processing the low frequency partial image of the original input image IMG, for example, the brightness of the outdoor image portion OUTD is dimmed, while the brightness of the indoor image portion IND is dimmed, as shown by the density of dots in fig. 8; b1', R ', OUTD ', IND ', P1' are images obtained by processing B1 and R, OUTD, IND, P1, respectively. Fig. 9 is a comparative diagram of the image adjustment operation of the high frequency part without additionally employing the image adjustment of the other range parameter and the image adjustment operation of the high frequency part with additionally employing the image adjustment of the other range parameter, as shown in the left side of fig. 9, the shadow image S1 'corresponding to the ladder image P1' is blurred, while the result of the image adjustment of the high frequency and low frequency parts simultaneously is shown in the right side of fig. 9, the shadow image S1 "corresponding to the ladder image P1" is clearer and sharper, that is, the detail level is improved, and the implementation of the present invention can also greatly improve the image detail such as the grid or the railing in the picture when the brightness adjustment of the bright and dark area is performed.

The foregoing description is only of the preferred embodiments of the invention, and all changes and modifications that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Claims

1. An image processing apparatus comprising:

a processing circuit for receiving an input image, comprising:

a high-frequency part processing unit, configured to perform a high-frequency part image adjustment operation according to a high-frequency information of the image of at least one pixel unit of the input image, and improve image details of the image of the at least one pixel unit; and

a low-frequency part processing unit, configured to perform a low-frequency part image adjustment operation according to low-frequency information of the image of the at least one pixel unit of the input image, and adjust a regional brightness value of the image of the at least one pixel unit;

wherein the processing circuit generates an output image according to the input image, the low frequency partial image adjustment operation, and the high frequency partial image adjustment operation.

2. The image processing apparatus according to claim 1, wherein the low-frequency information corresponds to the regional luminance value of the image of the at least one pixel unit, the regional luminance value being an average luminance value of the image of the at least one pixel unit and images of surrounding pixel units, and a curve equation employed by the low-frequency part processing unit is a curve equation of a local color correction algorithm, or a curve equation of a nonlinear enhancement algorithm, or a Smoothstep curve equation; the parameter z of the nonlinear enhancement algorithm is adjustable.

3. The image processing device according to claim 1, wherein the high-frequency information corresponds to an image reflectivity of the image of the at least one pixel unit, and a curve equation adopted by the high-frequency part processing unit is a curve equation of a local gamma correction operation algorithm or an S-type curve equation; the parameter gamma of the partial gamma correction operation algorithm is adjustable.

4. The image processing device of claim 1, further comprising:

a storage unit;

the processing circuit equally divides the range of pixel values of the image of the at least one pixel unit into N sections, equally divides the range of area brightness values of the image of the at least one pixel unit into M sections, generates (n+1) × (m+1) output pixel values according to (n+1) segmentation point values of the N sections, (m+1) segmentation point values of the M sections, a curve equation adopted by the high-frequency part processing unit and a corresponding high-frequency parameter value and a curve equation adopted by the low-frequency part processing unit and a corresponding low-frequency parameter value to generate a two-dimensional lookup table, and stores the generated two-dimensional lookup table in the storage unit.

5. The image processing apparatus according to claim 1, wherein when an incoming input pixel is received, the processing circuit generates a region luminance value corresponding to the input pixel, and finds a plurality of segmentation point values in the two-dimensional lookup table according to the value of the input pixel and the region luminance value of the input pixel, and performs a bilinear interpolation operation according to the value of the input pixel, the region luminance value of the input pixel, and the segmentation point values to generate an output pixel value.

6. An image processing method, comprising:

receiving an input image;

using a high-frequency part processing unit to perform a high-frequency part image adjustment operation according to high-frequency information of the image of at least one pixel unit of the input image, and improving image details of the image of the at least one pixel unit;

using a low frequency part processing unit to perform a low frequency part image adjusting operation according to low frequency information of the image of the at least one pixel unit of the input image, and adjusting a regional brightness value of the image of the at least one pixel unit; and

an output image is generated according to the input image, the low frequency partial image adjustment operation, and the high frequency partial image adjustment operation.

7. The image processing method according to claim 6, wherein the low-frequency information corresponds to the regional luminance value of the image of the at least one pixel unit, the regional luminance value being an average luminance value of the image of the at least one pixel unit and images of surrounding pixel units, and a curve equation system adopted by the low-frequency part processing unit is a curve equation of a local color correction algorithm, or a curve equation of a nonlinear enhancement algorithm, or a Smoothstep curve equation; the parameter z of the nonlinear enhancement algorithm is adjustable.

8. The image processing method according to claim 6, wherein the high-frequency information corresponds to an image reflectivity of the image of the at least one pixel unit, and a curve equation system adopted by the high-frequency part processing unit is a curve equation of a local gamma correction operation algorithm or an S-type curve equation; the parameter gamma of the partial gamma correction operation algorithm is adjustable.

9. The image processing method according to claim 6, further comprising:

providing a storage unit;

equally dividing a range of pixel values of the image of the at least one pixel unit into N segments;

equally dividing a range of region brightness values of the image of the at least one pixel unit into M sections;

generating (n+1) × (m+1) output pixel values according to the (n+1) segmentation point values of the N segments, the (m+1) segmentation point values of the M segments, a curve equation and a corresponding high frequency parameter value adopted by the high frequency part processing unit, and a curve equation and a corresponding low frequency parameter value adopted by the low frequency part processing unit to generate a two-dimensional lookup table; and

the generated two-dimensional lookup table is stored in the storage unit.

10. The image processing method according to claim 6, further comprising:

when an input pixel is received, a regional brightness value corresponding to the input pixel is generated, a plurality of segmentation point values are found in the two-dimensional lookup table according to the value of the input pixel and the regional brightness value, and a bilinear interpolation operation is performed according to the value of the input pixel, the regional brightness value of the input pixel and the segmentation point values to generate an output pixel value.