JP2004357086A - Encoding method and encoder of animation - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve a problem that many unused disk regions are generated when data is stored in a storage device in the unit of GOV with the fixed number of frames since activity of a video is greatly different depending on time and place when surveillance monitoring is targeted in the case of encoding animation data. <P>SOLUTION: The number of frames in a random access unit is varied according to data quantity and encoding control is performed so that multiples of the data quantity of the random access unit approach prescribed data quantity of the storage device. Thus, the unused regions of the disk are reduced and use efficiency of the storage device is enhanced. In addition, retrieval time in random access processing is stabilized and throughput of data transfer processing is enhanced. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention belongs to a moving image data encoding technique.
[0002]
[Prior art]
2. Description of the Related Art In a video surveillance system that stores and manages encoded data obtained by encoding moving images in a storage device, data storage processing is performed in consideration of data access efficiency. In such a system, in response to a request from a user, a necessary data chunk is extracted from the storage device to the video distribution server and data distribution processing is performed. At this time, the amount of data to be extracted from the storage device includes a disk array. There is an optimum value for the system in terms of the basic data size, the logical size of the disk, and the cache size of the video distribution server. In this specification, this optimum data amount is defined as a specified storage data amount, and an area on the disk that is divided by the specified storage data amount is defined as a storage data area.
When searching for necessary image data in a storage device, first, a storage data area including necessary data is searched, and then, encoded data in the storage data area is decoded to search for required image data. Take steps. Therefore, in such a system, the first frame of each storage data area is coded data of a frame that can be randomly accessed.
[0003]
There are roughly two types of video coding methods, which are called intra coding and inter coding, respectively. Inter-encoding utilizes previously encoded frames for prediction processing in order to use inter-frame correlation. Intra coding is a coding method using only intra-frame correlation. A frame encoded using only intra coding is called an intra frame, and a frame encoded using inter coding is called an inter frame. Therefore, the frames that can be randomly accessed are only intra frames that can be reproduced only with the encoded data of the frame.
[0004]
FIG. 5 shows a basic data structure of MPEG-4 video encoding as an example. The VOS header is profile / level information that determines the applicable range of MPEG-4 video products, the VO header is version information that determines the data structure of video encoding, and the VOL header is image size, encoding bit rate, frame memory size, application tools, etc. Information. The GOV header contains time information. The VOP is encoded data of each frame, and the VOP immediately after the GOV header is always an intra frame. Since the VOS header, VO header, VOL header, and GOV header all start with a 32-bit unique word, they can be easily searched. End code of VOS indicating the end of the sequence is also a 32-bit unique word. These unique words start with 23 “0” s and 1 “1”, and have a structure in which 2 bytes of data following the 24 bits indicate the type of delimiter.
As described above, the encoded data is stored in the storage data area in a randomly accessible data unit (hereinafter, referred to as a random access unit) starting from an intra frame such as a GOV composed of a GOV header and a plurality of VOPs. And implements random access processing.
[0005]
[Problems to be solved by the invention]
For the purpose of communication or broadcasting, from the viewpoint of decoding processing such as buffer control, the number of frames in the random access unit is fixed, and the number of frames in the random access unit is almost constant so that the code amount in the random access unit is almost constant. Perform data distribution. However, for the purpose of monitoring and monitoring, the image quality is important, and the activity of the video greatly differs depending on the time and place. Therefore, if the number of frames in the random access unit is fixed and data is stored in the storage device in the random access unit, a large amount of unused disk space occurs due to a mismatch between the prescribed storage data amount and the data amount in the random access unit, Disk usage efficiency is reduced. For example, an unused disk area occurs as shown in data 601 in FIG. 6 (relationship between the prescribed storage data amount and the random access data amount). Such a phenomenon is likely to occur particularly at midnight when there is little movement in the video. Further, when data of a random access unit straddles two storage data areas like the data 602, a large unused disk area is generated, and the disk use efficiency is reduced. Such a phenomenon is likely to occur when the motion in the video is large.
[0006]
Also, since the time required for the search process in the random access unit depends on the amount of data in the random access unit, if the number of frames in the random access unit is fixed, the time required for the random access process is not stable, This leads to a decrease in service quality.
[0007]
Furthermore, when the number of frames in the random access unit is fixed, the data amount in the storage data area varies, so that the efficiency of using the communication band when distributing data to the decoding device is reduced.
[0008]
[Means for Solving the Problems]
The number of frames in the random access unit is changed according to the data amount, and encoding control is performed so that a multiple of the data amount of the random access unit approaches the prescribed storage data amount of the storage device. More specifically, if the sum of the estimated code amount of the next input image and the total amount of unit data is larger than the prescribed storage data amount, after initializing the unit data total amount, the code amount of the input image is reduced to the prescribed storage data amount. The input image is intra-coded so as to be smaller, and the code amount is added to the unit data total amount. If the sum of the estimated code amount of the next input image and the total amount of unit data is smaller than the prescribed storage data amount, the input image is interpolated so that the sum of the code amount of the input image and the total amount of unit data becomes smaller than the prescribed storage data amount. Encode.
[0009]
BEST MODE FOR CARRYING OUT THE INVENTION
As described in the related art, in a network monitoring system including a storage device, data distribution processing from the storage device to the decoding device is performed for each storage data area. Therefore, the first frame of the encoded data stored in each storage data area needs to be an intra frame. In a coding apparatus that performs coding control (fixed coding rate) so that the coding rate does not fluctuate, intra-frames are generated at regular intervals of a certain number of frames as periodic processing, and one cycle is set as a random access unit. As an exceptional process, an intra-frame may be generated at the time of a scene change (corresponding to a surveillance video when a camera is operated or a camera is switched). This is because the amount of encoded data can be reduced. However, in surveillance applications, the quality of each image in a video scene needs to be high quality, and the code amount of each frame greatly differs between a time when inter-frame fluctuation is large and a time when inter-frame fluctuation is small. . Therefore, after applying the encoding control method of inserting intra frames at a fixed frame number interval, and recording encoded data in the storage device so that the first frame of the storage data area is intra-coded, the The rate of occurrence of unused areas in the disk increases, and the disk utilization efficiency decreases.
[0010]
Therefore, in the present invention, a data amount that is a guide is set for each random access unit, and encoding processing is performed so as to approach the set data amount. Then, the target data amount of the random access unit is set so that a multiple thereof becomes the prescribed storage data amount. As a result, the number of frames in the random access unit becomes variable, but the data amount in the random access unit is stabilized, and the value of each storage data amount becomes a value close to the specified storage data amount. as a result,
1) The unused area of the storage device decreases, and the amount of data that can be stored increases.
2) The number of storage data areas required to accumulate a certain amount of data is reduced, and the average search time of the storage data area including the encoded data of the frame to be accessed is reduced.
3) Since the amount of encoded data between two intra frames is stable, the maximum search time in a random access unit and the response delay at the time of a search request are reduced (necessary because random access processing involves decoding processing). The time depends on the amount of encoded data),
4) Since the amount of codes stored in each storage data area is stabilized, the efficiency of data retrieval from the storage device to the distribution server is improved.
5) Since the amount of codes stored in each storage data area is stabilized, the communication band to the decoding device and the utilization efficiency of the data path in the data storage device including the storage device and the distribution server are improved.
And other effects.
[0011]
FIG. 2 shows the code amount control method of the present application, taking as an example a case where the target data amount of the random access unit and the prescribed storage data amount are made to match. When the next input image is input (process 801), the sum of the prescribed storage data amount D, the unit total data amount T, and the estimated code amount p is compared (process 802). Here, the unit total data amount T indicates the total code amount from the first frame recorded in one storage data area to the immediately preceding recording frame. The estimated code amount p is an estimated code amount when the next input image is inter-coded, and is a value estimated from the code amount of the immediately preceding frame, the current time, the camera angle, the activity of the current random access unit, and the like. is there. If D is larger than p + T, the input image is inter-coded so that the code amount e becomes smaller than DT (process 803). On the other hand, if D is smaller than p + T, the T value is first initialized to 0, and the encoding process for the current storage data area in the immediately preceding frame ends. Then, with the input image as the first frame of the new storage data area, intra coding is performed so that the code amount e becomes smaller than D (process 804). After the process 803 is completed, e is compared with DT (process 805). If e is larger than DT, the encoding process of process 803 is canceled and process 804 is performed. If e is smaller than DT in the processing 805, the value of T is updated (T = T + e) (processing 806). Also, after the processing 804 is completed, the value of T is updated (T = T + e) (processing 806). Thereafter, the processes 801 to 806 are repeated.
[0012]
In the above description, for simplicity, the target data amount in random access units and the prescribed storage data amount are used. However, the point of the present invention is that the first frame of each data storage area is an intra frame, and the amount of data stored in each data storage area is close to the prescribed storage data amount. Therefore, the number of intra frames included in each data storage area is not limited, and in an environment where response speed to random access is important, it is better to provide a plurality of intra frames in the data storage area to improve service quality. Good. For example, a method of setting n times (n is an integer of 2 or more) the target data amount of the random access unit so as to be the specified storage data amount, preparing a plurality of candidates for the target data amount, and determining a combination of the specified storage amounts. You may set so that it may become a data amount. In this case, the following processing is added before the processing 802. First, the total data amount A + p of the random access unit is compared with the target data amount B (process 807). At this time, if A + p is greater than B, A is initialized to 0 (step 808). The value of A + p is greater than B, and the current random access unit is not the last random access unit allocated to the data storage area (process 809), but the number of encoded frames in the random access unit If c is larger than the minimum number F of frames in the random access unit (process 812), the input image is set as an intra frame of the next random access unit, p is set as an estimated code amount of intra coding (process 810), and The value is set to 1 (step 811). Whether the value of A + P is larger than B and the current random access unit is the last random access unit allocated to the data storage area (process 809), or the value of c is smaller than F In this case (process 812), the value of c is increased by 1 (process 811). If the value of A + P is smaller than B in process 807, the value of c is increased by 1 (process 811). The process 812 has the effect of solving the problem that the coding efficiency is reduced in a scene where the change between frames is large. In the present invention, since the number of frames in the random access unit is controlled by the amount of encoded data, the number of frames in the random access unit is reduced in a scene having a large inter-frame variation. Will be higher. Therefore, when the number of frames in one random access unit is smaller than the prescribed value F, the next random access unit is synthesized, and the frequency of occurrence of intra frames in the time direction is reduced. With this processing, it is possible to avoid a decrease in coding efficiency. The value of F is set to a value smaller than the average number of frames in the random access unit. Further, in a system in which the data delivery process from the storage device to the decoding device is n times the prescribed storage data amount (n is an integer of 2 or more), the target data amount in random access units is set to a value larger than the prescribed storage data amount. Can be set to In this case, a value m times the target data amount (m is an integer of 2 or more and smaller than n) in the random access data unit is set to an n times value of the prescribed storage data amount, and the value of D is n times the prescribed storage data amount. Control as a value.
[0013]
In the above description, in the process 805, when e + T> D, the encoding process of the process 803 is canceled and the process 804 is performed. However, e is set to 0 (without encoding the input image, ) And processing 806 may be performed.
[0014]
Data 801 to data 803 in FIG. 8 show examples of MPEG-4 encoded data generated by the processing in FIG. 2 (the random access unit is GOV). As described above, the number of frames belonging to one random access unit varies depending on the activity of the video scene, the camera angle, the shooting time, and the like. However, the code amount of each random access unit is a value close to the prescribed storage data amount. The unused area of each storage data area is reduced as compared with the case of FIG.
FIG. 7 shows a configuration example of a network monitoring system including a storage device that performs the processing of FIG. In FIG. 7, the encoding device sequentially inputs images of each angle captured by a number of cameras (1a, 1b, 1c, 1d,...) To the encoding device 2. The encoding device 2 sequentially encodes the input video from each camera. For example, when three cameras are connected to the encoding device 2 and the encoding rate of each camera is 1 frame / sec, the encoding is performed at 3 frames / sec while shifting the input timing to the encoding device. I do. As another example, there is a case where the frame rate is set to 10 frames / sec, the video of the camera that inputs data to the encoding device is switched for several seconds, and the encoding is sequentially performed. 11 is a configuration example of encoded data when the data 701 in FIG. 10 includes three cameras. As shown in data 702 (encoded data of camera 1) in FIG. 10, the encoded data of each camera may be recorded as individual data in the storage device. When recording individually, the processing in FIG. 2 is also performed individually for each camera. In surveillance applications, camera number information and data feature information may be added to encoded data to improve search efficiency. Also in this case, the present invention can be applied by adding the data size of such information to the value of the code amount e in FIG. In addition, when there is room in the processing of the encoding device, encoded data of a plurality of bit rates and screen sizes may be generated for each input image in order to support decoding devices of different specifications. In this case, the present invention can be applied by individually applying the processing of FIG. 2 to the encoded data of each specification. The encoded data is distributed to the data storage device 3 via the network. Although FIG. 7 shows only one encoding monitoring apparatus, a plurality of monitoring apparatuses are connected via a network. In the data storage device 3, the receiving server 3a receives the encoded data of the monitoring video provided from each encoding monitoring device, and the code amount as shown in FIG. 8 becomes the prescribed storage data amount (setting method will be described later). After being divided into close and randomly accessible data units (consisting of one or more random access units), the data is stored in the storage data area of the storage device 3b. In some systems, the target data amount in random access units may be set to a value larger than the prescribed storage data amount.) At this time, for the storage data area where the VOL header of FIG. 5 does not exist before the first frame data in the storage data area, it is possible to add a VOS header, a VO header, and a VOL header before the GOV header to the receiving apparatus. Distribution processing becomes easy. However, it is necessary to add the data size of such information to the value of the code amount e. The camera number and time information (image size and bit rate, if necessary) of the data stored in each storage data area are managed by the distribution server 3c. In the case where the encoding device 2 performs the code amount control of FIG. 2 for each camera individually, first, as shown in FIG. 10, the input encoded data (data 701) of each camera is It is divided into encoded data (data 702, an example of data of camera 1). Then, each camera is divided into data units (consisting of one or more random access units) in which the code amount as shown in FIG. 8 is close to the prescribed storage data amount (setting method will be described later) and is randomly accessible. Is stored in the storage data area. When receiving a command from the monitor 5 via the network, the distribution server 3c searches for a storage data area including the corresponding time and the encoded data of the camera, and stores the data in the memory area in the distribution server for each storage data area. The data is read and distributed to the decryption device 4. At this time, the profile of the playback terminal on the monitor side is received at the same time as the command, and if the image size of the encoded data stored in the storage device does not match the specifications of the playback terminal, perform transcoding. To deliver. When encoded data corresponding to a plurality of terminal specifications is stored in the storage device by the processing of the encoding device, the encoded data of the optimal specification is searched and distributed. When coded data of a plurality of cameras is multiplexed and stored in the storage device 3b as in the data 701 in FIG. 10, only necessary data can be distributed based on a command from a supervisor. It becomes possible. For example, when only the data of camera 1 is requested, data 702 is delivered, when all of cameras 1 to 3 are requested, data 701 is delivered, and when only cameras 1 and 2 are requested, data 703 is delivered. .
In the above description, data distribution from the storage device to the decoding device is performed at the request of the supervisor, but the received encoded data may be distributed to the decoding device in real time. In this case, the data is transferred from the receiving server 3a to the distribution server 3b and then stored in the storage device 3b, or the data is directly distributed from the receiving server 3a to the decoding device and then stored in the storage device 3b.
[0015]
Further, in monitoring applications, long-term recording of several months is required, and there is a possibility that the disk capacity of the storage device becomes insufficient. On the other hand, as the time elapses, the access frequency of the coded data stored in the storage device from the supervisor decreases, and its importance decreases. Therefore, processing for reducing the amount of old data is performed. As a method,
1) performing transcoding (processing such as reducing the image size, reducing the frame rate, or reducing the image quality) on the encoded data in the storage device;
2) remove bidirectional prediction frames (encoded data not used for prediction of other frames and used in encoding schemes such as MPEG) in encoded data;
And the like, and the code amount is reduced. Specifically, the receiving server or the transmitting server extracts old data from the storage device, performs the above-described code amount reduction processing, and stores the data again in the storage device. The data in the storage data area where the original data is stored can be overwritten. Also in this case, if a code amount control method such as the process 2 in FIG. 2 is applied, the disk use efficiency is improved. Such a data amount reduction process is also effective when performing a backup process to another storage device data.
[0016]
1, the internal configuration of the encoding device 2 in FIG. 7 will be described. In this configuration, an extra frame memory for storing the reproduced image (locally decoded image) is prepared by the number of cameras connected to the encoding apparatus (twice the number of cameras when bidirectional prediction is used). Assign to camera. Then, control is performed so that the image stored in the frame memory corresponding to the camera that captured the input image is used as a reference image at the time of inter-frame prediction. Reduce the rate. At this time, as in the case of data 701 or data 703 in FIG. 10, when coded data of video captured by a plurality of cameras is multiplexed and distributed, the switching information of the reference image is notified to the decoding side.
[0017]
In the basic moving image encoding process, one frame of a moving image includes one luminance signal (Y signal: 2001) and two color difference signals (Cr signal: 2002, Cb signal: 2003), and the image size of the color difference signal is 信号 of the luminance signal both vertically and horizontally. At the time of encoding, first, the input image 200 is divided into small blocks as shown in FIG. This small block is called a macroblock. FIG. 13 shows the structure of a macroblock. The macro block is composed of one Y signal block 2101 of 16 × 16 pixels, and a Cr signal block 2102 and a Cb signal block 2103 of 8 × 8 pixels that spatially match the Y signal block 2101. The Y signal block may be further divided into four 8 × 8 pixel blocks (2101-1, 2101-2, 2101-3, and 2101-4) for processing. The divided macroblocks are encoded by any of an intra-encoding method and an inter-encoding method.
[0018]
In the intra coding, the input macroblock image 201 is input to the DCT converter 203 for each of six 8 × 8 pixel blocks (2101-1, 2101-2, 2101-3, 2101-4, 2002, 2003), and 64 Are converted into DCT coefficients. Each DCT coefficient is controlled by a quantization parameter (a value that determines the accuracy of quantization, which is determined by the control unit 301, in which the moving range is 1 to 31 in MPEG-4, and the condition determined by the processing in FIG. 2 is satisfied). ) Is quantized by the quantizer 204. The quantized DCT coefficients are passed to a multiplexer 206 and encoded. At this time, the quantization parameter is also passed to the multiplexer 206 and encoded. The quantized DCT coefficients are decoded into an input block image by an inverse quantizer 207 and an inverse DCT unit 208 of the local decoder 220, and are combined in a frame memory 210. The local decoder 220 needs to have the ability to create the same decoded image on the decoding side. The image stored in the frame memory 210 is used for inter-frame prediction in the time direction.
[0019]
The frame memory 210 stores a decoded image of the input image to be encoded. The reference image memory 316 is provided with the same number of frame memories as the number of camera positions connected to the encoding device (when performing bidirectional prediction, two frames are required for each camera). Each frame memory in the reference image memory 316 has a one-to-one correspondence with each camera, and stores a reference image for the corresponding camera. The frame data in the frame memory 210 and the frame data in the reference image memory 316 are substantially managed without distinction, and when the camera that captures the image input to the encoding device is switched, the frame data in the frame memory 210 is changed. Is exchanged with the pointer of the memory in which the frame data corresponding to the camera before switching in the reference image memory 316 is stored in the reference image memory 316.
[0020]
In the inter encoding, first, a motion compensation process is performed by the motion compensator 211 between the input macroblock image 201 and the locally decoded image of the previous frame in the frame memory 210 corresponding to the input image. Motion compensation refers to a portion similar to the content of a target macroblock from a locally decoded image (reference image) of a previous frame (generally, the absolute value of a prediction error signal in a luminance signal block is compared with the search range of the previous frame). This is a compression technique in the time direction for searching for a portion having a small sum of values) and encoding the amount of motion (motion vector). FIG. 4 shows a processing structure of motion compensation. FIG. 4 is a diagram showing a prediction block 55 and a motion vector 56 on a reference image 53 for a search range 57 for a luminance signal block 52 of a current frame 51 surrounded by a thick frame. The motion vector 56 indicates the amount of movement from the block 54 (broken line) on the reference image that spatially corresponds to the same position as the thick frame block of the current frame to the predicted block 55 area on the reference image. (The motion vector length for the color difference signal is half of the luminance signal and is not encoded.) Normally, the local decoded image in the frame memory 210 is provided to the motion compensator as a reference image. However, at the time of encoding a frame immediately after camera switching, the switch 317 is switched to the reference image memory 316, and the reference image memory 316 is referred to. An image is provided. The procedure for selecting a reference image corresponding to the input image from the reference image memory 316 is as follows. When the camera that captures the image input to the encoding device is switched, camera switch information 313 including the current camera number is input to the control unit 301 from the camera system. In response to this, the control unit 301 notifies the switch 317 to the switch 317 and the camera number information 315 to the reference image memory 316. Thereby, the reference image at the time of motion compensation is switched to the reference image of the camera corresponding to the camera number information 315 in the reference image memory 316. The camera number information 315 needs to be notified to the decoding side. As a method, a method of combining the encoded data with the video encoded data and sending it is conceivable. The combination 2000 of the unique word and the camera number information which cannot be generated in the encoded data as shown in FIG. 9 may be combined before the video data of the frame in which the camera switching has occurred. The data size of the data 2000 needs to be added to the code amount e in FIG. The motion vector 212 detected by such motion compensation is encoded by the multiplexer 206. The predicted macroblock image 213 extracted from the reference image on the frame memory by the motion compensation is subjected to difference processing between the input macroblock image 201 of the current frame and the differentiator 202 to generate a differential macroblock image. Is done. The difference macroblock image is input to the DCT unit 203 for each of the six 8 × 8 pixel blocks (2101-1, 2101-2, 2101-3, 2101-4, 2002, and 2003) shown in FIG. It is converted into 64 DCT coefficients. Each DCT coefficient is supplied to the quantizer 204 in accordance with a quantization parameter (a value that determines the precision of quantization, which is a moving range of 1 to 31 in MPEG-4, and that satisfies the condition determined by the processing in FIG. 2). , And is passed to the multiplexer 206 together with the quantization parameter, and is coded. Also in the case of predictive coding, the quantized DCT coefficients are decoded into a difference macroblock image by the inverse quantizer 207 and the inverse DCT unit 208 of the local decoder 220, and added to the predicted macroblock image by the adder 209. After that, the image is synthesized with the frame memory 210.
[0021]
The determination between the intra coding (INTRA) and the predictive coding (INTER) is performed by the INTRA / INTER determination unit 214 in MB units. In general, the determination is made using INTER as the evaluation value as the sum of the absolute values of the prediction errors in the luminance signal block, and INTRA as the evaluation value as the sum of the absolute values of the differences from the average value in the luminance signal block. In the prediction method of the inter-encoding, in addition to the forward prediction in which a predicted macroblock image is generated using information of a temporally past frame, a predicted macroblock image is predicted by using information of a temporally future frame. And bidirectional prediction in which a predicted macroblock image is generated using temporally past and future frame information. In an encoding device that uses backward prediction or bidirectional prediction, it is necessary to prepare two reference images for each camera. In addition, although details are omitted in this specification, prediction within a frame is usually used also in intra coding. In the case of intra prediction, a locally decoded pixel of an image being coded, coded DCT coefficients, and the like are used for prediction. Such features of intra prediction do not affect the processing procedure of the present invention.
[0022]
Next, the setting process of the quantization parameter in the control unit 301 in FIG. 1 will be described. It is necessary for the control unit 301 to record a prescribed storage data amount before starting the encoding process. As a method of notifying the prescribed storage data amount, when the encoding device is set in advance, when the administrator of the encoding device inputs the information externally and changes the setting of the encoding device, the data storage device 3 is connected via the network. May be transmitted from the server. The prescribed storage data amount is a value that depends on the configuration of the data storage device, and does not usually change. However, when a disk exchange or the like of the storage device occurs, the data needs to be updated.
When the next input image is input, the control unit 301 estimates the estimated value p of FIG. 2 from the bit amount information 310 of the previous frame obtained from the multiplexing unit 206 and the fluctuation of the activity of the input image, and the like. To start. Then, when the process proceeds to the process 803 or 804, the coding type of the frame (intra frame, inter frame) and the target code amount of the input image (DT in process 803, estimated code amount of intra coding in process 804) ) Is determined. The control unit 301 controls the quantization parameter so that the code amount is close to the target code amount and smaller than the target code amount within a range where the image quality does not deteriorate, and performs the encoding process of the input image. . Thereafter, the processing 805 and the processing 806 are performed with the actual code amount (the processing 804 is also performed depending on the condition).
[0023]
The internal configuration of the decoding device 4 in FIG. 7 will be described with reference to FIGS. FIG. 11 shows a configuration of a decoding device that reproduces encoded data of an image captured by one camera like data 702 in FIG. 8. FIG. 3 shows data 701 in addition to data 702 in FIG. This is a configuration of a decoding device that can also reproduce data obtained by multiplexing encoded data of images captured by two or more cameras, such as data and data 703.
[0024]
In FIG. 11, first, the coded data input by the decoding unit 501 is analyzed and converted from binary code to meaningful decoded information. Then, the motion vector information and the prediction mode information (INTRA / INTER determination) are distributed to the motion compensator 504, and the quantized DCT coefficient information is distributed to the inverse quantizer 502. If the prediction mode of the analyzed macroblock is intra coding, the decoded quantized DCT coefficient information is dequantized by the inverse quantizer 502 and the inverse DCT unit 503 for each 8 × 8 pixel block. Inverse DCT processing is performed to reproduce a macroblock image. When the prediction mode of the macroblock is inter coding, first, the motion compensator 504 generates a predicted macroblock image. Specifically, a predicted macroblock image is extracted from the frame memory 507 storing the decoded image of the previous frame according to the motion amount of the motion vector information. Next, the coded data relating to the prediction error signal is subjected to inverse quantization and inverse DCT processing for each 8 × 8 pixel block in an inverse quantizer 502 and an inverse DCT unit 503 to reproduce a differential macroblock image. Then, the predicted macroblock image and the difference macroblock image are added by the adder 505 to reproduce the macroblock image. The reproduced macro block image is combined with the decoded frame image by the combiner 506. The decoded frame image is stored in the frame memory 507 for prediction of the next frame.
[0025]
In FIG. 3, it is necessary to control the reference image input to the motion compensator 504 to be the same image as the encoding side, and the switch 509 is controlled according to the decoding camera number information. As in the encoding device of FIG. 1, in addition to the frame memory 507 for storing the reproduced image, the reference image memory 508 has the same number of camera positions as the number of camera positions set in the encoding device. Double) Frame memory is available. The decoded image is stored in the frame memory 507. Each frame memory of the reference image memory 508 stores a reference image corresponding to each installed camera in the encoding device. The frame data of the frame memory 507 and the frame data of the reference image memory 508 are managed substantially without distinction. When the encoding / decoding unit 501 receives the camera number information, the frame data of the frame memory 507 is stored. The pointer of the stored memory is exchanged with the pointer of the memory in which the reference image of the camera corresponding to the image decoded last in the reference image memory 508 is stored. When the decoding unit 501 decodes the decoded camera number information, a switch command 510 is notified to the switch 509, and the input path to the motion compensator 504 is switched to the reference image memory 508 side. At the same time, the camera number information 511 is notified to the reference image memory 508. The reference image memory 508 provides the reference image corresponding to the camera number information 511 to the motion compensator 504. The switch 509 is switched to the frame memory 507 when the decoding of one frame of encoded data is completed.
[0026]
The application of the present invention does not depend on the configuration in the storage device. The encoding method of the present invention is applicable to a hard disk, a disk array, a magnetic tape, and an optical storage disk medium. Can be implemented. The present invention is characterized by an encoding method that takes into account a prescribed storage data amount determined by the performance of the storage device and the distribution server, and the prescribed storage data amount exists regardless of the configuration in the storage device. Elements that determine the prescribed storage data amount include the disk size of the disk array, the disk sector size, the logical data size of the disk, the sector size of the media, the logical data size of the media, and the cache size of the distribution server.
[0027]
The encoding method of the present invention can be applied to a single camera system having no multiple cameras as shown in FIG.
[0028]
The encoding method of the present invention can be applied to a system that records encoded data in a storage device, and is not limited to a monitoring system. For example, the present invention can be applied to a video distribution server that stores encoded data and distributes video on demand.
[0029]
【The invention's effect】
The unused area of the disk is reduced, and the utilization efficiency of the storage device is improved. Further, the search time in the random access process is stabilized, and the throughput of the data transfer process is improved.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a configuration of an encoding device according to the present invention.
FIG. 2 is a diagram illustrating a code amount control process according to the present invention.
FIG. 3 is a diagram illustrating a configuration of a decoding device that decodes encoded data generated according to the present invention.
FIG. 4 is a diagram illustrating the principle of motion compensation.
FIG. 5 is a diagram showing an overall configuration of a video encoded bit stream.
FIG. 6 is a diagram illustrating the relationship between the amount of encoded data in a random access unit generated by a conventional encoding method and a prescribed storage data amount.
FIG. 7 is a diagram illustrating an overall configuration of a network video monitoring system including a storage device.
FIG. 8 is a diagram illustrating the relationship between the amount of encoded data in a random access unit generated by the encoding method of the present invention and the prescribed amount of stored data.
FIG. 9 is a diagram illustrating a format of camera information.
FIG. 10 is a diagram illustrating an example of video data generated by the encoding method of the present invention.
FIG. 11 is a diagram illustrating another configuration of a decoding device that decodes encoded data generated according to the present invention.
FIG. 12 is a diagram illustrating macroblock division in video encoding.
FIG. 13 is a diagram illustrating a configuration of a macroblock in video encoding.
[Explanation of symbols]
Reference numeral 200: input image, 201: input macroblock image, 202: difference unit, 203: DCT processing unit, 204: quantization unit, 206: multiplexing unit, 207, 502: inverse quantization unit, 208, 503: inverse DCT 209, 505: adder, 210, 507: frame memory, 211, 504: motion compensator, 214: INTRA / INTER determination unit, 300: MB division processing unit, 301: control unit, 317, 509: switch, 313: camera switch information, 314, 510: switching instruction, 315, 511: camera number information, 316, 508: reference image memory, 501: code decoding unit, 506: combining unit, 513: display device.

Claims (3)

A moving image encoding device including a storage device, wherein the encoding device has a control unit for controlling a code amount, and the control unit defines a sum of an estimated code amount of the next input image and a total amount of unit data in advance. If it is larger than the prescribed storage data amount, the total amount of unit data is initialized, and the input image is intra-encoded so that the code amount of the next input image is smaller than the prescribed unit storage data amount, and the code amount is united. If the sum of the estimated code amount of the next input image and the total amount of unit data is smaller than the specified unit storage data amount, the sum of the code amount of the next input image and the total amount of unit data is The input image is intra-coded or inter-coded so as to be smaller than the prescribed unit storage data amount, and the code amount is added to the unit data total amount. The encoding apparatus according to claim 1, further comprising a path for inputting the prescribed storage data amount, and having a function of updating a prescribed value of the prescribed storage data amount. The coding apparatus according to claim 1 or 2 is a moving picture coding apparatus included in a surveillance system having a camera switching function by a switcher, and is usually provided with a path for inputting camera switch information. In addition to the frame memory for the reference image, it has a plurality of additional frame memories for storing the local decoded image. The additional frame memory has one-to-one correspondence with each monitoring location. It has a function of switching a reference image from an image stored in a normal frame memory to an image stored in an additional frame memory.

Images