JP5031419B2 - Encoder - Google Patents

Description

  The present invention relates to an encoding device that encodes, for example, an image of a turning camera.

With the spread of networks and the digitalization of recording media, for example, even in camera systems whose main purpose is surveillance, images are digitized and compressed, distributed to remote locations, or recorded on digital media that can be recorded for a long time. It is supposed to be.
The compressed image data used for distribution, storage, etc. is preferably transmitted in a transmission path with a limited bandwidth or recorded on a finite storage medium, so that the data amount is preferably constant. Therefore, when the image is compressed, a process called code amount control is performed in order to keep the data amount constant.

  Processing called code amount control includes a method called a 1-pass format and a method called a 2-pass format. In the 1-pass format, based on the fact that moving images have a certain degree of identity in the time direction, the encoding parameter of the next frame is determined based on the encoding result of the previous frame, and the code amount control is performed. is there.

  The 2-pass format is a method of controlling the code amount by at least two processing passes for image data. In the first processing pass, information on the feature amount and complexity of the image to be compressed is collected. Before the second processing pass, an encoding parameter for approaching the target code amount is determined based on these pieces of information. For example, in JPEG, the quantization table is optimized. In the second pass, the same image is encoded with newly determined encoding parameters.

Conventionally, for example, there is Patent Document 1 as a code amount control method in MPEG, and Patent Document 2 as a code amount control method in JPEG.
Specifically, in Patent Document 1, as a first processing pass, an input moving image signal is first divided into scenes composed of at least one frame that is temporally continuous, and a statistical feature amount is calculated for each scene. The scene content is estimated based on this statistical feature, and if the buffer can have a certain margin for the target bit rate as the second processing pass, encoding is performed according to the scene content. The bit rate is allocated within the upper and lower limits set by the user, and an efficient coding parameter is determined so that the average bit rate satisfies the bit rate specified in advance, and the input image signal is determined using this coding parameter. Encode.

JP 2001-245303 A (paragraph 0012) Japanese Patent Laid-Open No. 9-130802 (paragraph 0012)

  The conventional encoding device is configured as described above, and the two-pass format has the feature that the code amount can be made constant with high precision, but the cost increases due to complicated processing and increased buffer usage. There was a problem of becoming a factor.

  The present invention has been made to solve the above-described problems, and it is possible to make the code amount constant with high accuracy, simplify the processing, suppress the buffer usage amount, and reduce the cost increase factor. An object of the present invention is to obtain an encoding device that can be used.

Coding apparatus according to the present invention includes a quantization unit for quantizing input image images from the camera, and a plurality of quantization tables, each suited to the turning operation start and turning operation end and, when the turning operation start of the camera In addition, a quantization table suitable for starting the turning operation is selected from a plurality of quantization tables as a quantization table used in the quantization unit, and when the turning operation of the camera is finished, it is suitable for completing the turning operation from the plurality of quantization tables. painting Bei a code controller you select a quantization table using a quantization table by the quantization unit, the plurality of quantization tables, quantum code amount is output as the quantization result of the quantization unit is constant also of the a storing of characteristic values.

  According to the present invention, there is an effect that the code amount is made constant with high accuracy, the processing is simplified, the amount of use of the buffer is suppressed, and the cost increase factor is reduced.

An embodiment of the present invention will be described below.
FIG. 1 is a configuration diagram of a revolving camera system according to Embodiment 1 of the present invention. As shown in FIG. 1, digitized image data distributed from a revolving camera 101 is transmitted via a network 102. The operation PC 103 monitors using the image data, and the recorder device 104 records the image data.

FIG. 2 is a block diagram of an encoding apparatus according to Embodiment 1 of the present invention.
The encoding device is in the revolving camera 101, and the revolving camera 101 includes a camera unit 1 (camera), an encoding unit 10, and a revolving unit 30. The encoding unit 10 includes a memory 11, a DCT unit 12, a quantization table (first quantization table) 13, a quantization unit 14, a variable length encoding unit 15, a buffer 16, and a scale factor determination. Unit 17, code control unit 18, quantization table (third quantization table) 19, and quantization table (second quantization table) 20. The pan motor 32 and the tilt motor 33 are configured.

  Here, the memory 11 temporarily records an input image which is uncompressed image data from the camera unit 1, and the DCT unit 12 performs a discrete cosine transform (DCT) on the image input from the memory 11, and a quantization unit 14. Quantizes the data after DCT input from the DCT unit 12 using the quantization table 13. The variable length encoding unit 15 performs variable length encoding on the quantized data input from the quantization unit 14, and the buffer 16 transmits the encoded data input from the variable length encoding unit 15 to the outside.

  The scale factor determination unit 17 determines the next image based on the code amount output as the quantization result of the previously input image so that the code amount output as the quantization result of the input image from the camera becomes constant. A scale factor for calculating a quantization table to be used for quantization is determined.

  The quantization table 13 stores a quantization characteristic value that makes a code amount output as a quantization result of an input image from the camera unit 1 constant. The quantization characteristic value is a value used for quantization, and is set individually.

  The sign control unit 18 is provided with information on the state of turning control from the turning control unit 31 of the turning unit 30 during the turning operation of the camera unit 1, and the camera unit is selected from the plurality of quantization tables 19 based on the information. The quantization table 19 corresponding to the turning type and turning speed of 1 is selected, and the selected quantization table 19 is set in the quantization table 13. Further, the code control unit 18 sets the quantization table 20 corresponding to the preset operation in the quantization table 13 at the time of the preset operation of the turning operation of the camera unit 1 determined in advance.

  The quantization table 19 stores a quantization characteristic value that makes the code amount output as a quantization result of the input image from the camera unit 1 constant corresponding to the type of rotation and the rotation speed of the camera unit 1. The quantization table 20 has a constant code amount output as a quantization result of the input image from the camera unit 1 in accordance with a predetermined preset operation of the turning operation of the camera unit 1. It is a table which memorize | stores the quantization characteristic value. The quantization table 19 and the quantization table 20 are stored in a nonvolatile memory or the like.

  The turning unit 30 controls turning of the camera unit 1. When a turning control command is input from the outside, the turning control unit 31 analyzes the turning control command and operates the pan motor 32 and the tilt motor 33. The sign control unit 18 is notified of the start of the turning motion and information related to the turning motion. The pan motor 32 and the tilt motor 33 rotate the camera unit 1 in accordance with instructions from the turning control unit 31.

Next, the operation will be described.
When an image is input to the encoding unit 10 in the camera unit 1, the memory 11 temporarily records the input image, the DCT unit 12 performs discrete cosine transform on the image input from the memory 11, and the quantization unit 14 includes the DCT unit 12. The data after DCT input from Quantize is quantized using the quantization table 13, the variable length encoding unit 15 performs variable length encoding on the quantized data input from the quantization unit 14, and the buffer 16 stores the variable length code. The encoded data input from the encoding unit 15 is transmitted from the buffer 16 to the operation PC 103 and the recorder device 104 via the network 102.
Then, the quantization unit 14 outputs the code amount output as the quantization result to the scale factor determination unit 17, and the scale factor determination unit 17 inputs the input image from the camera based on the code amount input from the quantization unit 14. The scale factor for calculating the quantization table used for the quantization of the next image is determined so that the amount of code output as the quantization result of is constant. The quantization table 13 stores the quantization characteristic value based on the scale factor determined by the scale factor determination unit 17 and outputs the quantization characteristic value to the quantization unit 14. The quantization unit 14 is used as a quantization table 13 used in the next quantization.

  Here, when a turning control command for turning the camera unit 1 is input from the outside, the turning control unit 31 analyzes the input turning control command and operates the pan motor 32 and the tilt motor 33 to start the turning operation. At the same time, the turning control unit 31 notifies the code control unit 18 of the start of the turning operation and information related to the turning operation. The information related to the turning operation includes information such as the type of turning (pan operation, tilt operation, preset operation, etc.) and the turning speed.

The code control unit 18 that has obtained the information about the turning motion selects and calls the quantization table 19 suitable for the turning motion from the plurality of quantization tables 19 and sets the quantization table 19 in the quantization table 13. Therefore, at the start of the turning operation, the panning motor 32 and the tilting motor 33 are operated, so that the resolution and the like affecting the code amount of the image input from the camera unit 1 to the encoding unit 10 change significantly from the previous image. However, since the selected quantization table 19 is set in the quantization table 13 when quantized by the quantization unit 14 via the memory 11 and the DCT unit 12, the conventional two-pass format is not used. The code amount output as a quantization result is kept constant.
In addition, a quantization table most suitable for the turning speed or the like is previously registered in the quantization table 20 in conjunction with the preset operation, and the code control unit 18 is most suitable for the subject in conjunction with the preset number during the preset operation. 20 can be called.

Further, when the turning operation is finished, the turning control unit 31 stops the pan motor 32 and the tilt motor 33 and simultaneously notifies the code control unit 18 of information on the stop of the turning operation. The code control unit 18 that has obtained the information about the turning stop selects and calls the quantization table 19 suitable when the turn is not made from the plurality of quantization tables 19, and sets the quantization table 19 in the quantization table 13. In addition, since the subject after the turning motion is determined during the preset operation, a quantization table most suitable for the subject is registered in advance in the quantization table 20 in conjunction with the preset number, and the code control unit 18 stores the quantization table 20. Can be called.
Therefore, at the end of the turning motion, the resolution of the image input from the camera unit 1 to the encoding unit 10 is greatly changed compared to the previous image, but the selected quantization table 19 is set in the quantization table 13. Therefore, the code amount is kept constant rather than the conventional two-pass format.

FIG. 3 is an image of the pan operation of the camera unit 1 according to Embodiment 1 of the present invention, and is displayed as an image 301 before pan operation, an image 302 during pan operation, and an image 303 after pan operation. The pan operation is an operation for moving the camera unit 1 in the horizontal direction.
The pre-pan operation image 301 is a high-resolution image in which the subject is clear, and the in-pan operation image 302 is a diagram when the camera unit 1 performs a pan operation. The after-image is generated in the horizontal direction and the contour is not clear. The image becomes low. The post-pan operation image 303 is a diagram when the revolving camera finishes the pan operation thereafter, and is an image with a high resolution like the pre-pan operation image 301.
FIG. 3 shows an example of an image in the pan operation. However, the same applies to an operation of moving the camera unit 1 in the vertical direction, that is, a so-called tilt operation. The result is a high image, a low resolution image, and a high resolution image. In addition to the pan operation and tilt operation, the types of turning include a preset operation for moving the camera unit 1 to a pre-registered angle of view.

FIG. 4 is a correspondence diagram between the code amount output as the quantization result of the input image from the camera unit 1 and time, and FIG. 4A is a correspondence diagram between the code amount and time in the conventional one-pass format. FIG. 4B is a correspondence diagram between code amount and time in the first embodiment.
An example of a conventional one-pass format of the code amount output as a quantization result of an input image obtained by a series of turning operations will be described with reference to FIG. In the state before the turning operation, there is no significant difference between the previous image and the next image in the resolution that affects the code amount. Therefore, even if the quantization table 13 is calculated from the scale factor obtained from the quantization result of the previous image and used for the quantization of the next image, the actual code amount 401 becomes a value close to the target code amount 402. Next, when the camera unit 1 starts a turning motion, the quantization table 13 is calculated from the scale factor obtained from the previous image with high resolution (for example, 301 in FIG. 3), and the next image with low resolution (for example, FIG. Therefore, the actual code amount 401 becomes smaller than the target code amount 402 as in the turning motion start 403.

  Furthermore, since a low-resolution image (for example, 302 in FIG. 3) continues during the turning operation, there is no significant difference in resolution and the like that affect the code amount between the previous image and the next image, and the actual code amount 401 Is a value close to the target code amount 402, but when the camera unit 1 finishes the turning operation, the quantization table 13 is calculated from the scale factor obtained from the previous image (for example, 302 in FIG. 3) having a low resolution. Since it is calculated and used for quantization of the next image with high resolution (for example, 303 in FIG. 3), the actual code amount 401 becomes larger than the target code amount 402 as in the turning operation end 404.

  An example of a code amount output as a result of quantization of an input image obtained by a series of turning operations in Embodiment 1 of the present invention is as shown in FIG. Also in the turning operation start 413 and the turning operation end 414, the selected quantization table 19 is set in the quantization table 13, so the actual code amount 411 becomes a value close to the target code amount 412.

  The encoding unit 10 according to Embodiment 1 performs a turning operation on an image having a large difference between the previous image and the next image at the start of turning or at the end of turning, instead of the quantization table 13 obtained from the previous image. Quantization is performed using a quantization table 19 tailored to the above, and the code amount can be kept constant. Therefore, when distributing to a network, network traffic can be kept constant, and the distribution bandwidth is limited. Even when the recording is performed, frame dropping does not occur, and it is possible to prevent a shortage of recording media due to a change in data size during recording.

FIG. 5 is a flowchart showing the program operation of the code control unit 18 according to the first embodiment of the present invention.
After starting in step ST1, the sign control unit 18 determines whether or not the turning operation is started based on the information from the turning control unit 31 in step ST2. If the turning operation does not start, the code control unit 18 returns to step ST2 again to turn the turning operation. If there is a start, the process proceeds to step ST3. In step ST3, the code control unit 18 analyzes information related to the turning operation (turning type (pan, tilt, preset), speed, etc.), and in step ST4, the type and turning of the camera unit 1 stored in advance are stored. The quantization table 19 suitable for the turning motion is read out from the plurality of quantization tables 19 corresponding to the speed. For example, if the turning speed is fast, the amount of afterimage on the screen increases. Therefore, select the quantization table A that can take a large amount of code even for an image with low resolution. If the turning speed is slow, the amount of afterimage on the screen is small. The quantization table C that can be obtained with a smaller code amount than the quantization table A may be selected. Since the quantization table 19 read in step ST4 is set in the quantization table 13 in step ST5, fluctuations in the code amount at the start of the turning motion can be suppressed.

  During the turning operation, the scale factor determination unit 17 determines the scale factor from the quantization result of the previous image as usual, multiplies the quantization table 13 by the scale factor to generate a new quantization table 13, and the quantization unit 14 quantizes the next image with the newly generated quantization table 13. In step ST6, the sign control unit 18 determines whether or not the turning operation end information has been obtained from the turning control unit 31, and proceeds to step 7 when the turning operation is finished, and step when the turning operation is not finished. Return to ST6.

  Next, the code control unit 18 analyzes information related to the end of the turning motion in step ST7, and reads the quantization table 19 suitable for the end of the turning motion from the plurality of quantization tables 19 in step ST8. Since the quantization table 19 read in step ST8 is set in the quantization table 13 in step ST9, fluctuations in the code amount at the end of the turning motion can be suppressed.

  As described above, according to the first embodiment, the encoding unit 10 according to the present invention stores the quantization characteristic value that makes the code amount output as the quantization result of the input image from the camera unit 1 constant. Corresponding to the quantization table 13 and the rotation type and the rotation speed of the camera unit 1, the quantization characteristic value at which the code amount output as the quantization result of the input image from the camera unit 1 becomes constant When the plurality of quantization tables 19 stored and the turning operation of the camera unit 1 are selected, the quantization table 19 corresponding to the type and turning speed of the turning of the camera is selected from the plurality of quantization tables 19 and selected. By providing the code control unit 18 that sets the quantization table 19 in the quantization table 13, the code amount is made constant with high accuracy, the processing is simplified, the amount of use of the buffer is suppressed, and the cost increase factor The effect is to reduce.

  Further, according to the first embodiment, the quantization characteristic value in which the code amount output as the quantization result of the input image from the camera unit 1 is constant corresponding to the preset preset operation of the turning operation of the camera. And a code control unit 18 that sets the quantization table 20 corresponding to the preset operation in the quantization table 13 at the time of the preset operation of the camera unit 1, the code amount is fixed with high accuracy. In addition, the process is simplified, the amount of buffer used is suppressed, and the cost increase factor is reduced.

  Conventionally, processing delay occurs because it requires time to pass two processing passes. However, according to the first embodiment, the processing delay can be prevented as compared with two processing passes. Play.

  Further, according to the first embodiment, there is an effect that the code amount can be made constant with high accuracy as compared with the conventional one-pass processing device.

Embodiment 2. FIG.
FIG. 6 is a block diagram of an encoding apparatus according to Embodiment 2 of the present invention.
A revolving camera 101 according to the second embodiment includes a camera unit 1 (camera), an encoding unit 10, and a revolving unit 30. The encoding unit 10 includes a memory 11, a DCT unit 12, a quantization table 613, a quantization unit 14, a variable length encoding unit 15, a buffer 16, a scale factor determination unit 17, and a code control unit 618. The turning unit 30 includes a turning control unit 31, a pan motor 32, and a tilt motor 33.

  In the first embodiment, the quantization table 13 itself is changed in order to optimize the quantization table 13 at the start or end of the turning motion. In Embodiment 2, by controlling the scale factor of the scale factor determination unit 17 to be multiplied when the quantization table 613 for the next image is calculated, the code amount at the start and end of the turning motion The fluctuation of the is suppressed.

  When a turning control command is input from the outside to the turning unit 30, the turning control unit 31 starts the turning operation by operating the pan motor 32 and the tilt motor 33 according to the turning control command. Notify information about start and turn motion. The code control unit 618 that has obtained information on the start of the turning operation and the turning operation is a scale factor that makes the code amount output as the quantization result of the input image constant based on the type of turning and the turning speed of the camera unit 1. Is instructed to determine the scale factor determination unit 17. The smaller the value of the scale factor, the larger the amount of code, and the larger the value, the smaller the amount of code.For example, if an input image with a high turning speed and a large amount of afterimage is expected, the scale factor should be reduced. It ’s fine. The quantization table 613 stores a quantization characteristic value based on the scale factor determined by the scale factor determination unit 17. At the start of the turning operation, the image input from the camera unit 1 to the encoding unit 10 changes more greatly than the previous image, but since the quantization table 613 is set corresponding to the turning operation, the quantization unit When quantization is performed at 14, the code amount output as the quantization result is kept constant. At the end of the turning operation, the code amount is kept constant by the same operation.

  Since the quantization table 13 itself is changed in the first embodiment, the quantization table 13 is changed regardless of the subject situation that has been taken by the camera until then. In the second embodiment, the scale factor is changed and multiplied to the quantization table 613 used up to the previous image. Therefore, the quantization table 613 taking into consideration the subject situation that the camera unit 1 has taken until then, Become. Therefore, it can be expected that the actual code amount becomes closer to the target code amount. In the second embodiment, since it is not necessary to store the quantization table 19 and the quantization table 20, the quantization table 19 and the quantization table 20 are unnecessary, and the H / W cost can be reduced.

  As described above, according to the second embodiment, the encoding apparatus 10 according to the present invention first inputs the code amount output as the quantization result of the input image from the camera unit 1 and the like so as to be constant. The scale factor determination unit 17 determines a scale factor for calculating a quantization table used for quantization of the next image based on the code amount output as the quantization result of the image, and the scale factor determination unit determines the scale factor. The quantization table 613 that stores the quantization characteristic values based on the scale factor and the rotation result of the camera unit 1 are output as the quantization result of the input image based on the type of rotation and the rotation speed of the camera unit 1. A code control unit 618 for instructing the scale factor determination unit 17 to determine a scale factor for which the code amount to be constant is determined. While a constant, processing simplify, suppressing the use amount of the buffer, an effect that reduces the cost increase.

1 is a configuration diagram of a revolving camera system according to Embodiment 1 of the present invention. FIG. It is a block diagram of the encoding apparatus by Embodiment 1 of this invention. It is an image of the pan operation | movement by Embodiment 1 of this invention. FIG. 6 is a correspondence diagram between a code amount output as a result of quantization of an input image from the camera unit according to Embodiment 1 of the present invention and time. It is a flowchart figure about operation | movement of the program of the code | symbol control part by Embodiment 1 of this invention. It is a block diagram of the encoding apparatus by Embodiment 2 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Camera part, 10 encoding part, 11 Memory, 12 DCT part, 13,613 Quantization table, 14 Quantization part, 15 Variable length encoding part, 16 Buffer, 17 Scale factor determination part, 18,618 Code control part , 19, 20 Quantization table, 30 swivel unit, 31 swivel control unit, 32 pan motor, 33 tilt motor, 101 swivel camera, 102 network, 103 operation PC, 104 recorder device, 301 image before pan operation, 302 pan operation in progress Image, 303 Image after pan operation, 401, 411 Actual code amount, 402, 412 Target code amount, 403, 413 Start of turning operation, 404, 414 End of turning operation.

Claims (2)

A quantization unit for quantizing input image images from the camera,
A plurality of quantization tables each suitable for the start and end of the turning motion of the camera ;
In the case of starting the turning operation of the camera, a quantization table suitable for starting the turning operation is selected from the plurality of quantization tables as a quantization table used in the quantization unit, and in the case of the end of the turning operation of the camera, e Bei a selection to that code controller a quantization table suitable for pivoting operation end from said plurality of quantization tables as the quantization table used in the quantization unit,
The plurality of quantization tables store quantization characteristic values at which a code amount output as a quantization result of the quantization unit is constant.
It marks Goka apparatus. It said plurality of quantization tables includes a multiple quantization table suitable for the type and the turning speed of the swivel operation,
The code control unit, when the turning operation start of the camera, using a quantization table suitable for the type and the turning speed of the swivel operation from the multiple quantization table in the quantization unit It is selected as the quantization table
Encoding apparatus according to claim 1, wherein the this.

Images