JP2006080832A - Image transcoder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image transcoder capable of enhancing a compression rate of image coded data while maintaining subject quality of an image. <P>SOLUTION: The image transcoder includes: an image input means 21 for receiving coded image data 22; a quality setting means 23 for outputting region quality information 25 comprising information for setting at least temporal image quality by each region on the basis of target region information 24 comprising information of an image region and information for denoting a notice degree by each image region; a coding conversion means 26 for converting the received image data 22 into image coded data 28 with image quality on the basis of the region quality information 25 by each image region, and an image output means 27 for outputting the image coded data 28. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image transcoder for converting encoded image data.

  An image transcoder is a device that receives encoded image data and converts the encoded image data into image encoded data having a different bit rate or encoding method. For example, the image transcoder shown in Non-Patent Document 1 does not convert the entire area of the image data with a uniform spatial image quality, but the spatial image quality of an image area with a high degree of attention in the image data. An image quality change type conversion is performed to increase the image quality and decrease the spatial image quality of the background, which is the other image area. Here, the spatial image quality includes a level at which a high-frequency component of an image is stored and an image resolution.

  Further, in the image transcoder shown in Non-Patent Document 1, region cropping type conversion is performed in which the background is deleted and only an image region having a high degree of attention of image data is converted with high spatial image quality. By performing such image quality change type and area cropping type conversion, the amount of image data to be transmitted is reduced and the bit rate is reduced. The region can be observed in more detail.

Naoki Kuwabara, 3 others, “Video Gateway for Surveillance Applications”, IEICE Technical Report, September 2003, IE2003-57, pp. 13-18

  In security monitoring and road monitoring, image data areas where moving objects such as pedestrians and vehicles have high priority, and background areas with little movement have low priority. However, when image quality change type conversion is performed by the image transcoder shown in Non-Patent Document 1, only the spatial image quality is converted in the background area to reduce the amount of data, so that there is a limit to the compression of image data. In addition, when the region cropping type conversion is performed by the image transcoder shown in Non-Patent Document 1, the background region is not displayed and the surrounding state is not known at all. There was a problem that the subjective quality of

  Accordingly, the present invention has been made to solve the above-described problems, and is an image that increases the compression rate of the encoded image data according to the degree of attention of each region while maintaining the subjective quality of the image. The purpose is to provide a transcoder.

  The solving means according to the present invention is based on attention area information including image input means for inputting encoded image data, information on the image area, and information indicating the degree of attention for each image area. Quality setting means for outputting area quality information consisting of information for setting at least temporal image quality, and converting the input image data into image encoded data having image quality based on the area quality information for each image area Encoding conversion means for performing image output, and image output means for outputting encoded image data.

  Since the image transcoder described in the present invention changes not only the spatial image quality but also the temporal image quality based on the attention degree of the region, the input image data is converted into image encoded data. While maintaining the subjective quality of the image, it is possible to increase the compression rate of the encoded image data and achieve a smaller output bit rate.

(Embodiment 1)
FIG. 1 is a system configuration diagram of an image monitoring system to which the image transcoder according to the first embodiment is applied. In this image monitoring system, a plurality of cameras 1 are installed to monitor roads, rivers, buildings, airports, etc., and an analog image signal from each camera is connected to a motion encoder JPEG (Joint Photographic) by a network encoder 2 provided in each camera. It is encoded into encoded image data such as Expert Group) or MPEG-2 (Motion Picture Expert Group phase 2) and distributed to the disaster prevention network 3. The distributed image data is decoded and displayed by the image display terminal 4 installed in the monitoring room, and is recorded in the image storage / distribution server 5. Usually, the images observed by security guards are required to be of high quality, so image encoded data with an output bit rate of several Mbps or higher is adopted, and the disaster prevention network 3 also uses a broadband network of 100 Mbps or 1 Gbps class. The

  On the other hand, when the owner of a building views the site at the office terminal 9 in the office via the office network 8 whose transmission bandwidth is limited, or on the mobile terminal 11 or the personal computer 12 at home via the narrow-band public network 10. When viewing the site, image data of several Mbps or more cannot be transmitted as it is. Therefore, in the image monitoring system, the network transcoder 6 is provided, and the input high bit rate image encoded data is encoded and converted (transcoded) into the low bit rate image encoded data, and then passed through the gateway 7. And distributed to the office network 8.

  The image transcoder according to the present embodiment is mounted on the network transcoder 6. There are various image transcoder mounting methods including a software mounting method and a hardware mounting method. For example, the network transcoder 6 main body uses a normal personal computer having a network function, and the image transcoder is mounted on the personal computer as software. In the present invention, there is no limitation on the method of mounting the image transcoder.

  FIG. 2 is a block diagram of the image transcoder according to the present embodiment. First, in the image transcoder shown in FIG. 2, the image data 22 encoded by the network encoder 2 is input to the image input means 21, and the image input means 21 passes the image data 22 to the encoding conversion means 26. On the other hand, the quality setting means 23 receives the attention area information 24 and sets temporal and spatial image quality for each area according to the attention level of each area. Then, the quality setting unit 23 outputs the set image quality information as region quality information 25 to the encoding conversion unit 26. The encoding conversion means 26 performs encoding conversion processing on the image data 22 using the area quality information 25 as a parameter.

  The encoding conversion means 26 encodes and converts it into image encoded data 28 having an image quality and encoding format different from the image data 22 based on the given region quality information 25. In particular, with regard to image quality, in an area with a high degree of attention (attention area), encoding conversion is performed so that the spatial image quality is high and the temporal image quality is also high. That is, the attention area has a resolution sufficient for a still image and a high frame rate. On the other hand, in regions such as the background other than the region of interest, coding conversion is performed so that both temporal and spatial image quality is low. The image output means 27 receives the encoded image data 28 converted by the encoding conversion means 26 and outputs it to the gateway 7 via the transmission means of the network transcoder 6 (not shown).

  The network transcoder 6 receives the image data 22 distributed in the stream format from the network encoder 2, performs sequential encoding conversion, and distributes it to the network. The attention area information 24 is defined in advance for each camera, and the contents are described in the setting file. FIG. 2 also shows a schematic diagram of the image data 22 and the encoded image data 28. The image data 22 shown in FIG. 2 shows data for seven frames with uniform spatial image quality. In the encoded image data 28 shown in FIG. 2, the attention area in the center of the image has high spatial image quality and high temporal image quality (data for seven frames). It is shown that the background other than the area has low spatial image quality and low temporal image quality (data for three frames of t0, t3, and t6).

  Next, the operation of the image transcoder according to the present embodiment will be described below. First, an example of attention area information 24 is shown in FIG. 3, an example of an image quality table is shown in FIG. 4, and an example of area quality information 25 is shown in FIG. In the attention area information 24 shown in FIG. 3, two areas of one attention area and a background area are defined, with the area ID = 1 indicating the attention area and the area ID = 2 indicating the background. Further, in the attention area information 24 shown in FIG. 3, the area ID = 1, coordinate information (X11, Y11), (X12, Y12) indicating the upper left and lower right of the attention area, attention degree “5” information, and the area The coordinate information indicating the background and the information of the attention degree “1” are stored in ID = 2.

  The quality setting means 23 reads the attention area information 24 shown in FIG. 3, and creates the area image quality information 25 shown in FIG. 5 with reference to the image quality table 32 shown in FIG. For example, the attention area of area ID = 1 has a degree of attention of “5”, so the temporal image quality is 15 fps, the spatial image quality is level 5, and the background area of area ID = 2 has the attention degree of “5”. Therefore, the temporal image quality is set to 5 fps, and the spatial image quality is set to level 1.

  Here, the temporal image quality is a frame rate (fps (frames per second)), and indicates how many frames of an image are formed per second. The encoding conversion means 25 realizes temporal image quality conversion by thinning out frames of the input image data 22. When encoding is performed within a frame such as JPEG or JPEG2000, only necessary frames are simply encoded and converted, and unnecessary frames are discarded. However, when encoding between frames such as MPEG-2 / 4, all B frames or all B frames and P frames are simply thinned out, or some frames are thinned out and motion vectors are calculated again. Various methods are generally known, such as a method to perform the above. Also in this embodiment, since a generally known method is used, detailed description of each method is omitted in this specification.

  On the other hand, spatial image quality indicates, for example, a level at which high-frequency components of an input image are stored, and is defined from level 5 for storing all to level 1 for storing only a DC component. The removal of high frequency components can be realized by changing the quantization step in the re-encoding of the encoding conversion.

  The image quality table 32 can be defined in various ways according to the purpose. For example, in FIG. 4A, both temporal image quality and spatial image quality are changed in accordance with the degree of attention, whereas in FIG. The image quality table 32 emphasizes the spatial image quality in which the temporal image quality is changed from the maximum 15 fps to 1/5 fps (1 frame per 5 seconds). FIG. 4C shows an image quality table 32 in which the temporal image quality is fixed at 15 fps and the temporal image quality is changed from level 5 to 1. The image quality table 32 in FIG. 4C is suitable for monitoring that emphasizes movement. As described above, the image transcoder according to the present embodiment can change various temporal and spatial image qualities according to the purpose.

  Further, the quality setting means 23 creates area quality information 25 shown in FIG. 5 from the attention area information 24 shown in FIG. 3 and the image quality table 32 shown in FIG. The area quality information 25 shown in FIG. 5 stores area ID, coordinates, attention level, temporal image quality, and spatial image quality information. For example, in the attention area of area ID = 1, the coordinates (X11, Y11), (X12, Y12), attention degree “5”, temporal image quality “15 fps”, and spatial image quality “level 5”. ing.

  Next, the operation of the image transcoder will be specifically described with reference to the screen examples of FIGS. FIG. 6 shows a screen example of the input image data 22 for six frames. FIG. 7 is a screen example compared with the present embodiment, in which the spatial image quality of only the background area 34 is lowered by the image transcoder. That is, in FIG. 7, for all image frames, a high spatial image quality is given to the attention area 33 in the lower half of the screen where the person is moving, and a low spatial image quality is given to the background area 34 such as a building. The conversion is given.

  FIG. 8 shows an example of a screen obtained by performing coding conversion using the image transcoder according to the present embodiment. FIG. 8A shows encoded image data 28 output from the image transcoder. Specifically, since the area quality information 25 shown in FIG. 5 is followed, the attention area 33 has the same temporal and spatial image quality (15 fps, level 5) as the input image data 22, and the background area 34 It is encoded and converted at a temporal and spatial image quality (5 fps, level 1) lower than the input image data 34. As can be seen from FIG. 8A, the image data 22 in the background region 34 is thinned out of the three frames.

  Next, the display module of the display terminal receives the encoded image data 28 that has been encoded and converted by the image transcoder, sequentially decodes and synthesizes it for each frame, and forms a screen example as shown in FIG. In other words, the first frame is data obtained by encoding the entire area of the image including the attention area 33 and the background area 34, and the display module decodes this, expands it in the display memory, and displays it on the display. Since the second and third frames are data in which only the attention area is encoded, the decoded partial image data is overwritten at the position of the attention area in the display memory. This is repeated from the fourth frame.

  As is clear from the comparison of the screen examples shown in FIGS. 7 and 8, in the present embodiment, the frame of the background area 34 can be thinned out to be converted into less image encoded data 28, and the background area When there is little movement of 34, the subjective quality of the almost same image can be obtained. Even if the background area 34 moves, if the area is not important as a monitoring target, the user can concentrate on observing the important area without being noticed by the extra movement.

  FIG. 9 is another screen example in the present embodiment. FIG. 9A shows encoded image data 28 output from the image transcoder. The temporal and spatial image quality of the attention area 33 shown in FIG. 9A is the same as that of FIG. 8A, but the temporal image quality of the background area 34 is 5 fps, and the spatial image quality is the same. “Level 5” is set, and the spatial image quality of the background region 34 is set high. That is, in the example shown in FIG. 9A, at least the temporal image quality of the background region 34 is lowered to improve the spatial image quality, and the subjective quality of the image is improved. In the example shown in FIG. 9A, the data amount increases compared to the example shown in FIG. FIG. 9B shows a composite display screen displayed on the display module. Since the spatial image quality of the attention area 33 and the background area 34 is the same, the attention area can be visually recognized. The range 33 is superimposed and displayed in a rectangular frame with a broken line.

  10 to 12 are examples of screens when two attention regions 33 are set. FIG. 10 shows an example of a screen compared with the present embodiment, in which only the background area 34 has a reduced spatial image quality. FIG. 11 shows a screen example when the temporal and spatial image quality of the attention area 33 is set to (15 fps, level 5) and the background area 34 is set to (5 fps, level 1) in the present embodiment. FIG. 12 shows a screen example when the image quality of the attention area 33 is set to (15 fps, level 5) and the background area 34 is set to (5 fps, level 5). FIG. 11A shows encoded image data 28 output from the image transcoder, and FIG. 11B shows a composite display screen displayed on the display module.

  FIG. 12 shows a screen example when the image quality of the attention area 33 is set to (15 fps, level 5) and the background area 34 is set to (5 fps, level 5). 12A shows the encoded image data 28 output from the image transcoder, and FIG. 12B shows a composite display screen displayed on the display module. Similar to the case of providing one region of interest 33 shown in FIGS. 7 to 9, when two regions of interest 33 are provided, the data is compressed more than the case where only the background region 34 reduces the spatial image quality. In addition, since the spatial image quality of the background region 34 is increased, the subjective quality of the image is improved.

  As described above, in the image transcoder described in the first embodiment, the quality setting unit 23 displays temporal and spatial image quality corresponding to the attention degree of each area based on the attention area information 24. Based on the area quality information, the coding conversion means 26 sets the temporal and spatial image quality in the area with a high degree of attention based on the area quality information, and the area with a low degree of attention such as the background is at least temporal. Since the encoding conversion is performed so that the image quality is lowered, the subjective quality of the image can be improved. In addition, there is an effect that it is possible to concentrate on observation of an important area without being limited by a time change of an unimportant area as a monitoring target.

  In the present embodiment, the encoding conversion timing of the encoded image data 28 has been described for the case of sequential conversion as streaming. However, the batch type that receives the image data 22 as a file and converts the data in a batch. Various timings depending on the application, such as conversion, can also be considered. In addition, the attention area information 25 has been described for a case in which the attention area information is set in advance for each camera. However, the attention area information 25 can be changed interactively by the observing user, or depending on the detection results of various sensors arranged at the shooting location. In some cases, it can be changed dynamically.

  Furthermore, as an example of changing the temporal image quality, the method of thinning out the frame of the input image data 22 and the method of changing the quantization step as an example of changing the spatial image quality have been described above. Is not limited to this. For example, there is a method of changing color information such as a luminance component and a color component to change the resolution of the image if the image quality is spatial. In the case of an encoding method such as JPEG2000, necessary portions of image resolution, frequency components, and color components can be combined for each region and extracted from high-quality image encoded data. If you want to change both temporal image quality and spatial image quality at the same time, use the motion compensation used in MPEG-2 / 4 etc. to change the quantization step of the prediction error according to the degree of attention And so on.

  The image quality table 32 defines the same temporal and spatial image quality for one degree of attention, but defines a plurality of sets of temporal and spatial image quality for one degree of attention. However, a method of selecting according to the content of the image may be used. For example, the image content is classified into a case where there is motion and a case where there is little motion. The former image (temporal image quality 15 fps, spatial image quality level 3), and the latter image (temporal image). The quality 7fps and the spatial image quality level 5) are defined, and the encoding / conversion means 26 selects an appropriate temporal / spatial image quality while analyzing the image content. Further, instead of the image quality table 32, the degree of attention is defined as a variable x, the temporal image quality and the spatial image quality are variables yt and ys, respectively, and defined as an equation of (yt, ys) = G (x). Also good.

(Embodiment 2)
In the present embodiment, the image transcoder according to the first embodiment has a configuration in which a function for dynamically changing the bit rate after coding conversion is added in accordance with network bandwidth fluctuations or user instructions. The image transcoder according to the present embodiment can also be applied to the image monitoring system shown in FIG.

  FIG. 13 is a block diagram showing a functional configuration of the image transcoder according to the present embodiment. The following description will focus on differences from the image transcoder according to Embodiment 1 shown in FIG. First, a bit rate input means 41 is newly provided in the image transcoder shown in FIG. The bit rate input means 41 receives an instruction to change the bit rate of the encoded image data 28 from an external module such as a GUI (Graphical User Interface) application of the display module or a bandwidth management manager, and sets a new target bit to the quality setting means 23. A rate value or a target data amount (hereinafter simply referred to as a target bit rate value) is passed.

  The quality setting means 23 resets the temporal and spatial image quality for each area of the image data 22 based on the attention area information 24 and the image data 22 so that the received new target bit rate value is obtained. The quality information 25 is changed. The encoding conversion means 26 receives the newly updated region quality information 25 and performs encoding conversion of the image data 22 according to the region quality information 25. For example, in the present embodiment, when the network traffic changes in time from medium to large to small, the 3 Mbps image data 22 is changed from 1 Mbps to 512 Kbps to 2 Mbps image encoded data 28 in accordance with the change. Convert it. FIG. 13 schematically shows how the bit rate of the input image data 28 is constant and the bit rate of the output image encoded data 28 changes.

  Bit rate control in the image transcoder according to the present embodiment will be described. First, when changing temporal image quality, the image transcoder starts decimation using the ratio Rin / Rout of the bit rate Rin of the image data 22 and the target bit rate Rout as the initial value of frame decimation. The image transcoder measures the latest bit rate after a certain period of time, and dynamically changes the thinning rate depending on whether the bit rate is higher or lower than the target bit rate.

  When changing the spatial image quality, the image transcoder uses, for example, a quantization step for each target bit rate set in advance based on the statistical properties of the image as an initial value within one frame of the image data 22. Requantize in units of macroblocks. The image transcoder dynamically adjusts the quantization step size depending on whether or not the accumulated data amount after re-quantization is likely to be higher than the target bit rate. The bit rate control according to the present embodiment uses a combination of the above temporal bit rate control and spatial bit rate control techniques. There are various variations on how to combine them. For example, as shown in FIG. 14 (a), an initial value of temporal image quality and an initial value of spatial image quality according to the target bit rate input by the bit rate input means 41 and the attention level of the area. Are set in the image quality table 32 in advance. The top level of the image quality table 32 indicates the target bit rate.

  The quality setting means 23 creates area quality information 25 corresponding to the target bit rate with reference to the image quality table 32 shown in FIG. FIG. 15 shows an example of the area quality information 25 when the target bit rate is 512 bps. Next, the encoding conversion means 26 performs both the temporal bit rate control and the spatial bit rate control using the initial value of the region quality information 25. The image quality table 32 shown in FIG. 14B is an example in the case where the temporal image quality is fixed for each target bit rate and only the spatial bit rate control is performed, and the image quality shown in FIG. In the table 32, the spatial image quality is fixed for each target bit rate, and only temporal bit rate control is performed.

  By changing the contents of the image quality table 32, various controls are possible depending on the purpose and application. If the statistical properties of the input image data 22 are considerably different for each camera, the camera is classified, and an image quality table 32 in which appropriate values are set for each type is prepared, and quality setting means 23 selects the quality setting means 23 according to the camera instructed to change by the bit rate input means 41.

  Next, the operation of the image transcoder according to the present embodiment will be described using the screen examples of FIGS. The screen examples in FIGS. 16 to 19 are examples in which the image data 22 in the screen example shown in FIG. 6 is encoded and converted by the image transcoder of the present embodiment. First, FIG. 16 to FIG. 18 show examples in which the bit rate is reduced by changing only the image quality of the background area 34 without changing the image quality of the attention area 33 as compared with the encoding conversion of FIG. Specifically, FIG. 16 shows a case where only the temporal image quality of the background region 34 is lowered. In FIG. 8, 2 out of 3 frames are thinned out. In FIG. 16, 4 out of 5 frames are thinned out. 16A shows the encoded image data 28 output from the image transcoder, and FIG. 16B shows a composite display screen displayed on the display module.

  FIG. 17 shows a case where both the temporal and spatial image quality of the background region 34 are lowered. In FIG. 17, 3 out of 4 frames are thinned out, and the spatial image quality of the background area 34 is reduced. FIG. 17A shows the encoded image data 28 output from the image transcoder, and FIG. 17B shows a composite display screen displayed on the display module. Further, FIG. 18 shows a case where only the spatial image quality of the background area 34 is lowered. 8A shows the encoded image data 28 output from the image transcoder, and FIG. 8B shows a composite display screen displayed on the display module.

  The spatial image quality of the background area 34 in the screen example shown in FIG. 18 is lower than the spatial image quality of the background area 34 in the screen example shown in FIG. In the screen examples shown in FIGS. 16 to 18, the level of spatial image quality is expressed by hatching, and the example in which finer hatching is applied indicates that the spatial image quality is lower. ing. Furthermore, the screen example shown in FIG. 19 improves the temporal image quality of the background region 34 and increases the bit rate compared to the screen example shown in FIG. In FIG. 19, one frame is thinned out of two frames. FIG. 19A shows the encoded image data 28 output from the image transcoder, and FIG. 19B shows a composite display screen displayed on the display module.

  As described above, according to the image transcoder according to the present embodiment, the quality setting is performed so that the bit rate input unit 41 becomes a new target bit rate value in response to the bit rate change instruction received from the external module. Since the area quality information 25 is changed by the means 23, an image can be distributed with a stable quality even in a network where there is a band limitation or fluctuation.

  The network transcoder 2 may be configured to distribute the changed bit rate and area quality information to the display side in synchronization with the encoded image data in accordance with the timing corresponding to the bit rate change. Thereby, the GUI application can display the image quality as supplementary information in accordance with the reproduction display of the image.

(Embodiment 3)
FIG. 20 is a system configuration diagram of an image monitoring system to which the image transcoder according to the present embodiment is applied. The image monitoring system shown in FIG. 20 differs from the image monitoring system shown in FIG. 1 in that the function of the image transcoder is implemented in the image storage / distribution server 5 without separately providing the network transcoder 2 as hardware. It is a point. The image storage / distribution server 5 receives the image stream distributed from the network encoder 2 and always endlessly records the latest image for a fixed time. The image storage / distribution server 5 records the event so that the latest image stream is not overwritten in an important time section of the recorded image stream, and moves to another recording area to store and record. Recording.

  In the image storage / distribution server 5 according to the present embodiment, when there is a distribution request from the office terminal 9 or the mobile terminal 11 to the image storage / distribution server, the image stream received from the network encoder 2 or the stored and recorded image The stream is encoded and converted by an image transcoder to reduce the bit rate, and is distributed via the gateway 7.

  Next, FIG. 21 shows a block diagram showing a functional configuration of the image transcoder according to the present embodiment. The difference between the image transcoder according to the present embodiment and the image transcoder shown in FIG. 2 is that the region-of-interest information generation unit 51 extracts a region to be noted from the image data 22, and information on the extracted region is obtained. This is used to generate attention area information 24. Specifically, for image monitoring, moving objects such as people and vehicles, landslides, and fire events are important. Must be identified as the background area.

  The attention area information generation means 51 shown in FIG. 21 decodes the image data 22 and extracts a moving body and an event occurrence area by image processing, and is defined in accordance with the extracted area information and preset rules. The attention area information 24 including information indicating the attention degree of each area is generated. A technique for tracking a moving object or detecting an event from the image data 22 is a conventionally known technique, and a technique for detecting a scene change or the like without decoding the image data 22 has also been announced. Therefore, detailed description is omitted in this specification.

  The attention area information 24 is updated by the movement of the object or the occurrence of an event, and each time the quality setting means 23 updates the area quality information 25 based on the updated attention area information 24. Then, the coding conversion means performs coding conversion of the image data 22 into the image coded data 28 according to the updated region quality information 25. The image transcoder according to the present embodiment is provided with the attention area information generation means 51, so that it is possible to automatically extract a moving object or an event occurrence area without human intervention, and a code corresponding to the content (content) of the image. Conversion can be realized.

  Next, the operation of the image transcoder according to the present embodiment will be described using the screen examples of FIGS. Each screen example is an example in which the image data 22 of the screen example shown in FIG. 6 is encoded and converted. First, FIG. 22 is an example of a screen compared with the present embodiment, in which an area including a pedestrian is encoded as an attention area 33 with high spatial image quality and a background area 34 is encoded with low spatial image quality. This is the case.

  Next, the screen example shown in FIG. 23 is a screen example that is encoded and converted by the image transcoder according to the present embodiment. FIG. 23A shows the encoded image data 28 output from the image transcoder, and FIG. 23B shows a combined display screen displayed on the display module. The first frame in FIG. 23A is composed of data obtained by encoding the attention area 33 including the pedestrian with high spatial image quality and data obtained by encoding the background area 34 with low spatial image quality. ing. However, in the second and third frames in FIG. 23A, data obtained by encoding the attention area 33 in the frame with high spatial image quality and the image that was the attention area 33 in the previous frame. The area is composed of data encoded with the same low spatial image quality as the background area 34. In the image transcoder according to the present embodiment, encoding conversion is performed by thinning out the back area 34 of the second and third frames, and the back area 34 of two frames out of three frames is thinned out.

  In the display module, the encoded image data of the screen example shown in FIG. 23B is decoded and displayed in the display memory. The encoded image data in the second and third frames is overwritten at the corresponding position of the encoded image data in the first frame. In the present embodiment, the same operation is repeated after the fourth frame. As a result, compared to the case of the screen example shown in FIG. 22, encoding conversion is performed into fewer image encoded data while maintaining the subjective quality of the same image.

  The screen example shown in FIG. 24 is a screen example of the image transcoder according to the present embodiment in which the spatial image quality of the background region 34 is set to the same level as the attention region 33, and is the screen example shown in FIG. Compared to the case, the subjective quality of the image is improved. 24A shows the encoded image data 28 output from the image transcoder, and FIG. 24B shows a composite display screen displayed on the display module. In the screen examples shown in FIG. 22 to FIG. 24, all the areas including the pedestrian are set as the attention area 33. The attention area 33 including the pedestrian moving in the right direction and the pedestrian moving in the left direction. And a region of interest 33 including.

  However, the screen examples shown in FIGS. 25 to 27 are screen examples when only the area including the pedestrian moving in the right direction is set as the attention area 33. 25 corresponds to FIG. 22, FIG. 26 corresponds to FIG. 23, and FIG. 27 corresponds to FIG. 24. FIG. 26 shows a temporal and spatial image of the background region 34 in the image transcoder according to the present embodiment. When the quality is set low, FIG. 27 shows a case where only the temporal image quality of the background area 34 is set low. Even if the image transcoder according to the present embodiment is set so as to be the screen example of FIG. 26 or FIG. 27, the subjective quality of the image is set as in the case of the screen example of FIG. 23 or FIG. The image data can be further compressed while maintaining. FIGS. 26A and 27A show the encoded image data 28 output from the image transcoder, and FIGS. 26B and 27B show the composition displayed on the display module. It is a display screen.

  As described above, according to the present embodiment, the attention area information generating unit 51 is configured to extract the moving object and the event occurrence area from the image data 22 and generate / update the attention area information 24. Therefore, it is possible to automatically realize the encoding conversion according to the content (content) of the image without human intervention.

  The image storage / distribution server 5 distributes the attention area information 24 and area quality information 25 to the display side in synchronization with the encoded image data 28 in accordance with the update timing of the attention area information 24. Also good. Accordingly, the GUI application can be configured to display the attention area and the image quality as supplementary information in accordance with the reproduction and display of the image. Also in this embodiment, the network transcoder 2 may be provided and the network transcoder 2 may be provided with an image transcoder. Further, in the first and second embodiments, the image transcoder may be provided in the image storage / distribution server 5 and the network transcoder 2 may not be provided, as in the present embodiment.

(Embodiment 4)
The image transcoder according to the present embodiment is also mounted on the image storage / distribution server in the image monitoring system of FIG. 20 as in the third embodiment. FIG. 28 shows a block diagram of the image transcoder according to the present embodiment. The difference between the image transcoder according to the present embodiment and the image transcoder according to the third embodiment is that the attention area information generating means 51 generates the background image data 61. The encoding conversion means 26 performs encoding conversion using the background image data 61 and the image data 22 and outputs image encoded data 28. The attention area information generation means 51 employs a background difference method that is generally used for moving object tracking and event detection, and can generate background image data 61 from the input image data 22.

  Next, the operation of the image transcoder according to the present embodiment will be described using the screen examples of FIGS. Each screen example is a screen example in the case where the region including only the pedestrian moving in the right direction is encoded and converted as the attention region 33 with respect to the image data of the screen example shown in FIG. In the first frame shown in FIG. 29 (a), the background image data 61 generated by the attention area information generating means 51 is encoded with a low spatial image quality, and the attention area 33 is encoded with a high spatial image quality. Yes. In the second and third frames shown in FIG. 29A, the attention area 33 in the frame is encoded and converted with high spatial image quality.

  In the display module, the encoded background image data 61 and the data of the attention area 33 are decoded, expanded in different areas of the working memory, copied to the display memory, and as shown in FIG. 29 (b). Display the composite screen. For the second and third frames, the background image data 61 already developed in the working memory in the first frame and the data of the attention area 33 of each frame are decoded and developed in the working memory. The data is copied to the display memory and displayed as a composite. The above operation is repeated after the fourth frame. FIG. 30 shows an example in which the background image data 61 generated by the attention area information generation means 51 is encoded with the same spatial image quality as the attention area 33. 30A shows the encoded image data 28 output from the image transcoder, and FIG. 30B shows a composite display screen displayed on the display module.

  In the present embodiment, as in the screen examples shown in FIGS. 26 and 27, the data of the image area that was the attention area 33 in the previous frame as shown in the third embodiment becomes unnecessary. The data amount of the encoded image data 28 can be reduced, and a smaller output bit rate can be achieved. In addition, since a pedestrian who moves to the left does not appear in the background image, it is possible to display an image in which other than the person or object of interest is visually excluded.

  As described above, according to the present embodiment, the attention area information generation unit 51 generates and updates the background image data 61 from the image data 22, and the encoding conversion unit 26 updates the background image data 61 and the image data 22. Therefore, the compression rate of the image data can be further improved while maintaining the subjective quality of the image. Furthermore, in this embodiment, unnecessary moving objects in the background area can be deleted, and an image optimal for monitoring the attention area can be displayed.

  In the present embodiment, the attention area information generating unit 51 generates the background image using the input image encoded data, but the present invention is not limited to this. For example, a configuration may be adopted in which a scene in which no moving object is present in advance when the camera is installed and the scene is given to the encoding conversion unit 26 as fixed background image data 61. The fixed background image data 61 is not limited to one, but a plurality of prepared background image data with different conditions such as time zone and weather may be prepared and selected according to the conditions of the encoded image data to be encoded and converted and supplied to the encoding conversion means. good.

  Further, the configuration is not limited to the configuration in which the background image data 61 is used only when the attention area information generation unit 51 generates and updates the attention area information 24. Even when the attention area information 24 is fixed in advance for each camera, the user interactively changes it, or dynamically changes the sensor signal or the like from the outside, the same effect as the present embodiment can be obtained by using the background image data 61. . Further, in the present embodiment, the network transcoder 2 may be provided, and the network transcoder 2 may be provided with the image transcoder.

(Embodiment 5)
Similarly to the third embodiment, the image transcoder of the present embodiment is also mounted on the image storage / distribution server in the image monitoring system of FIG. FIG. 31 and FIG. 32 are examples of screens after coding conversion is performed by the image transcoder of this embodiment. The image transcoder that is the image example of FIG. 31 has the same configuration as the image transcoder of the third embodiment that is the image example shown in FIG. However, unlike the image transcoder of the third embodiment, the image transcoder of the present embodiment is different from the image transcoder of the third embodiment in that the encoding / conversion means 26 performs the second and third data of the attention area 33 and the background area 34 in the first frame. The data of the attention area 33 in the frame is encoded and converted as still image data of one frame as shown in FIG.

  By performing encoding conversion as in the screen example shown in FIG. 31B, it is possible to observe a state (trajectory) of a pedestrian moving as a single still image during reproduction on the display module. Note that the same processing is repeated for the fourth and subsequent frames, and the image indicating the locus is updated and displayed. In addition, various variations are conceivable as a method of synthesizing the frames. For example, the three encoded image data are simply arranged in order, defined as encoded data as one frame, decoded in order in the display module and expanded in the display memory, and the data of the attention area 33 is placed at the corresponding position. Overwrite.

  As another method, a portion corresponding to the attention area 33 in the current frame is deleted from the image encoding data from the image encoded data of the previous frame, and the deleted portion is imaged in the attention area 33. Replace with encoded data. Specifically, in the case of JPEG encoding, an area is set for each macroblock, and is replaced with DCT data of the attention area 33 at the stage of DCT (Discrete Cosine Transform) data before performing Huffman encoding. Furthermore, when overwriting or replacing, it is possible to make the image of the attention area 33 translucent and to perform α composition superposition display.

  The image transcoder of the screen example shown in FIG. 32 has the same configuration as the image transcoder of the fourth embodiment, which is the screen example shown in FIG. FIG. 32A shows the encoded image data 28 output from the image transcoder, and FIG. 32B shows a composite display screen displayed on the display module. 32 is different from the image transcoder in the screen example of FIG. 31 in that the image showing the locus is generated using the background image data 61 and the image data 22; The background image data 61 and the attention area 33 are encoded and converted with the same spatial image quality.

  As described above, in the present embodiment, the encoded image data of the attention area 33 of each frame is synthesized as still image data of one frame with respect to a plurality of frames in which the encoding conversion unit 26 is continuous. The change of the attention area 33 at a certain time interval can be condensed and displayed in a single still image, and the state of the change can be grasped in a short time. In this embodiment, a network transcoder 2 may be provided and an image transcoder may be provided in the network transcoder 2.

(Embodiment 6)
Similarly to the third embodiment, the image transcoder according to the present embodiment is also mounted on the image storage / distribution server 5 in the image monitoring system of FIG. FIG. 33 shows a block diagram of the image transcoder according to the present embodiment. This image transcoder is obtained by adding the bit rate control function described in the second embodiment to the image transcoder of the fourth embodiment. In other words, in the present embodiment, the attention area 33 extracted by the attention area information generating means 51 changes with time in accordance with the appearance / disappearance of a pedestrian, and the area and the degree of attention of the attention area accordingly. Change. When the coding conversion is performed with a certain temporal and spatial image quality, the bit rate of the encoded image data 28 to be output also varies with respect to the change in the degree of attention.

  Therefore, in the present embodiment, when the attention area information 24 is updated, the quality setting means 26 makes the target bit rate value received by the bit rate input means 41 based on the degree of attention and the area of the attention area 33. The area quality information is updated and passed to the encoding conversion means 26. Based on the updated area quality information 25, the encoding conversion unit 26 performs encoding conversion processing so as to obtain a target bit rate value by the bit rate control function described in the second embodiment.

  Next, the operation of the image transcoder according to the present embodiment will be described using the screen examples of FIGS. The screen example of FIG. 34 shows the input image data 22, and one pedestrian is shown in the first four frames, and two pedestrians are shown in the next four frames. Assuming that the pedestrian is the attention area 33, the area of the attention area changes almost double in the first four image frames and the next four image frames. Therefore, if the given bit rate is used effectively and encoded and converted with high image quality in accordance with the content of the image, the temporal and spatial images of each region in the first 4 frames and the next 4 frames. You need to change the quality settings.

  In the screen example of FIG. 35, the temporal / spatial image quality of the attention area 33 is not changed, but only the temporal image quality of the background area 34 is changed. For example, in the first four frames, the frame rate of the background region 34 is ½ that of the input image, that is, the frame rate of the background region 34 is thinned out at a rate of one frame per two frames. Is thinned out to 1/4 of the input image, that is, at a rate of 1 frame to 4 frames.

  Conversely, in the screen example shown in FIG. 36, the temporal and spatial image quality of the attention area 33 is not changed, but only the spatial image quality of the background area 34 is changed. For example, the spatial image quality of the background area 34 is level 2 in the first four frames, but is level 1 in the next four frames. In FIG. 36, the level of spatial image quality is expressed by hatching, the first four frames are hatched by diagonal lines in one direction, and the next four frames are hatched by diagonal lines in both directions. In the screen example shown in FIG. 36, the spatial image quality of the background area 34 is lower than that in the screen example shown in FIG. 34, but this is suitable for the case where the background area 34 is not important but it is desired to observe the time change gently. Yes.

  As described above, in the present embodiment, when the attention area information 24 changes with the change of the image content, the quality setting means 23 updates the area quality information so that the target bit rate value is obtained, and the updated area quality is obtained. Since the encoding conversion means 26 performs encoding conversion based on the information, even if the number and area of the attention area 33 change as the image contents change, the limited quality of the image is high by effectively using the limited band. Encoding conversion can be realized.

  Also in this embodiment, the network transcoder 2 may be provided, and the network transcoder 2 may be provided with an image transcoder.

  In the above description, the input / output of the image transcoder may be quasi-moving image encoded data having a long frame interval. In addition, although it has been described that the same attention level is set for a plurality of attention areas 33, different attention levels are set for each area or for each type of area (for example, a pedestrian moving to the right and a pedestrian moving to the left). Also good.

1 is a system configuration diagram of an image monitoring system according to Embodiment 1 of the present invention. 1 is a block diagram of an image transcoder according to Embodiment 1 of the present invention. FIG. It is a figure which shows attention area information which concerns on Embodiment 1 of this invention. It is a figure which shows the image quality table which concerns on Embodiment 1 of this invention. It is a figure which shows the area | region quality information which concerns on Embodiment 1 of this invention. It is a figure which shows the example of a screen of the image data input into the image transcoder which concerns on Embodiment 1 of this invention. It is a figure which shows the example of a screen of the image coding data compared with the image transcoder which concerns on Embodiment 1 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 1 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 1 of this invention outputs. It is a figure which shows the example of a screen of the image coding data compared with the image transcoder which concerns on Embodiment 1 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 1 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 1 of this invention outputs. It is a block diagram of the image transcoder which concerns on Embodiment 2 of this invention. It is a figure which shows the image quality table which concerns on Embodiment 2 of this invention. It is a figure which shows the area | region quality information which concerns on Embodiment 2 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 2 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 2 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 2 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 2 of this invention outputs. It is a system block diagram of the image monitoring system which concerns on Embodiment 3 of this invention. It is a block diagram of the image transcoder which concerns on Embodiment 3 of this invention. It is a figure which shows the example of a screen of the image coding data compared with the image transcoder which concerns on Embodiment 3 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 3 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 3 of this invention outputs. It is a figure which shows the example of a screen of the image coding data compared with the image transcoder which concerns on Embodiment 3 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 3 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 3 of this invention outputs. It is a block diagram of the image transcoder which concerns on Embodiment 4 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 4 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 4 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 5 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 5 of this invention outputs. It is a block diagram of the image transcoder based on Embodiment 6 of this invention. It is a figure which shows the example of a screen of the image data input into the image transcoder which concerns on Embodiment 6 of this invention. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 6 of this invention outputs. It is a figure which shows the example of a screen of the image coding data which the image transcoder concerning Embodiment 6 of this invention outputs.

Explanation of symbols

1 camera, 2 network encoder, 3 disaster prevention network, 4 image display terminal, 5 image storage / distribution server, 6 network transcoder, 7 gateway, 8 office network, 9 office terminal, 10 public network, 11 mobile terminal, 12 PC, 21 image input means, 22 image data, 23 quality setting means, 24 attention area information, 25 area quality information, 26 encoding conversion means, 27 image output means, 28 image encoded data, 32 image quality table, 33 attention Area, 34 background area, 41 bit rate input means, 51 attention area information generation means, 61 background image data.

Claims (5)

Image input means for inputting encoded image data;
Quality of outputting region quality information consisting of information for setting at least temporal image quality for each image region based on attention region information consisting of information on the image region and information indicating the degree of attention for each image region Setting means;
Encoding conversion means for converting the input image data into image encoded data having image quality based on the area quality information for each image area;
An image transcoder comprising image output means for outputting the image encoded data. The encoding conversion means converts a plurality of continuous frames constituting the image encoded data into still image data of one frame including a continuation of at least a part of the image area in the continuous frames,
The image transcoder according to claim 1, wherein the image output means outputs the still image data as one still image or a time-series set of still images. A bit rate input means for inputting a target bit rate or a target data amount of the encoded image data output from the image output means to the quality setting means;
The said quality setting means updates the information regarding the image quality for every image area in the said area quality information based on the said target bit rate or the said target data amount, The Claim 1 or Claim 2 characterized by the above-mentioned. Image transcoder.   Based on the image data from the image input means, extract an attention image area, generate the attention area information using information on the attention image area, and output the attention area information to the quality setting means The image transcoder according to claim 1, further comprising a generation unit. The coding conversion means combines the image data of the image area of interest and background image data consisting only of a background image, and performs coding conversion to the image coding data based on the area quality information. The image transcoder according to any one of claims 1 to 4.

Images