JP3952926B2 - Video encoding apparatus and video encoding method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a video encoding device and a video encoding method, and more particularly to a video encoding device and a video encoding method for encoding a video signal having a scene change.
[0002]
[Prior art]
Conventionally, in supermarkets, convenience stores, pachinko parlors, etc., surveillance systems using cameras are used to prevent crimes such as shoplifting and to preserve evidence. In this monitoring system, in order to record a monitoring image from a camera, a long-time recording monitoring VTR called a time lapse VTR is used. Time-lapse VTRs are used not only for surveillance purposes for crime prevention and security, but also for work analysis, physical distribution records, and long-term experimental observation data recording in schools and laboratories.
[0003]
FIG. 3 shows a system configuration diagram of an example of the monitoring system. In the figure, video signals of moving images in a store to be monitored, which are photographed by four cameras 21a, 21b, 21c, and 21d, are switched by the switcher 22 every arbitrary time and supplied to the television monitor 23. It is displayed and monitored, and is supplied to the die muraps VTR 24. The time lapse VTR 24 records on the magnetic tape a time-series synthesized video signal obtained by switching a plurality of video signals in a short time by the switcher 22. By reproducing the recorded magnetic tape, the recorded contents can be analyzed in detail, or a specific reproduction screen can be taken and stored, and crime prevention, security monitoring and evidence preservation can be performed.
[0004]
When the alarm sensor 25 is called, the video signal supplied to the time lapse VTR 24 is switched by the switcher 22 to the video signal from the camera 21b in the vicinity of the alarm sensor 25, and the video signal is continuously recorded in the time lapse VTR 24. Perform alarm recording.
[0005]
However, since the time lapse VTR 24 uses a magnetic tape as a recording medium, the recording / reproducing head may be clogged or recording may not be possible due to tape entrainment. There may be a problem that the scene cannot be recorded.
[0006]
Therefore, in recent years, a video disk recorder (hereinafter referred to as VDR) that records a video signal on a hard disk or an optical disk has been increasingly employed in order to solve the problem of the time lapse VTR. Further, in order to efficiently record a moving image intermittently, image compression is performed on the image signal of the moving image to be recorded.
[0007]
Generally known image compression methods include a JPEG (Joint Photographic Experts Group) method and an MPEG (Moving Picture Experts Group) method (ISO / IEC DIS 13818-1 to 3). The JPEG method is an intra-frame compression method, in which one image is compressed into one block of data. On the other hand, the MPEG system is an inter-frame compression system, and is an intra frame (hereinafter referred to as an I picture) that is encoded using a spatial correlation within one frame when an image signal is encoded. Predicted frames (hereinafter referred to as P pictures) to be encoded using the correlation with the previous I picture, and interpolation (Interpolated) using the correlation between the I picture and the P picture in both directions ) The frame is divided into three frames (pictures) and processed (hereinafter referred to as B picture).
[0008]
In this MPEG system, a group consisting of a plurality of pictures from one I picture to a P or B picture immediately before the next I picture is called a GOP (Group of Picture). In general, in the case of the same image quality, the compression rate is higher in the MPEG method that performs inter-frame correlation than in the JPEG method that is intra-frame compression.
[0009]
By the way, the JPEG method is generally adopted as the monitoring VDR compression method described above. This is because in the monitoring VDR, since a signal obtained by switching a plurality of images in a short time is recorded, it is difficult to adopt the MPEG method which is a method for compressing continuous moving images.
[0010]
FIG. 4 is a block diagram showing an example of an image signal recording / reproducing apparatus for monitoring. The image signal recording / reproducing apparatus 31 for monitoring is a VDR using the JPEG method, and a standard television signal obtained by imaging a subject such as a monitoring target with a camera (not shown) is an A in the VDR 31. After being supplied to the / D conversion unit 32 and digitizing only the necessary frames, the JPEG compression unit 33 compresses the frames using the JPEG method. The compressed digital image signal (image data) is recorded on the disk device 34.
[0011]
The image data recorded on the disk device 34 is read from the disk device 34 during reproduction, decompressed by the JPEG decompression unit 35, and then supplied to the D / A conversion unit 36, where it is returned to an analog image signal. Is output. Normally, this output analog image signal is input to a television monitor (not shown), and an image is displayed on the screen.
[0012]
The series of recording / reproducing operations of the VDR 31 is performed by the central processing unit (CPU) 38 through the bus 39 for A / D conversion based on the recording / reproducing mode input by the operation unit 37 and the number of frames to be recorded. Of the blocks from the unit 32 to the D / A conversion unit 36, the necessary block is instructed.
[0013]
[Problems to be solved by the invention]
However, since the JPEG compression unit 33 is used as a video encoding device in the monitoring image signal recording / reproducing apparatus, not only a video signal obtained by switching a plurality of images in a short time but also a continuous video signal is encoded. Even when recording is performed, image compression is performed by the JPEG method, and there is a problem that the compression rate is lower and the recording time is shorter than the MPEG method. This is because if two types of compression methods are employed in one recording / reproducing apparatus, two types of compression circuits are required, which is disadvantageous in terms of cost and compatibility of recorded data.
[0014]
The present invention has been made in view of the above points, and can record a video signal or a continuous video signal obtained by switching a plurality of images in a short time such as a plurality of frames within a GOP time by using the MPEG system. It is an object to provide an encoding device and a video encoding method.
[0015]
[Means for Solving the Problems]
In order to achieve the above object, the encoding apparatus of the present invention encodes a video signal using a correlation between an I picture for encoding a video signal using a spatial correlation in one frame and a previous I picture. Encoding from one of the three pictures of the P picture, which is a prediction frame to be converted, and the B picture, which is an interpolation frame to be encoded using the correlation between both directions of the I picture and the P picture. In a video encoding apparatus that outputs an encoded signal in which a plurality of pictures up to a picture immediately before a picture are grouped as GOPs, a scene change detection unit that detects a scene change of the video signal, and a scene change detection unit scene change detection signal is judged that a scene change continuously when continuous plural frames in time of GOP, biopsy showing a scene change When the signal is not input, or a scene change continuously determining unit when the detection signal indicating the scene change is not input in a plurality of frames to determine the scene change nonconsecutive in time GOP, a scene change in the scene change continuously discriminating unit If the continuous is determined, all GOP composed of I picture, if it is determined discontinuous scene change in the scene change continuous determination unit, all three pictures the GOP or I-picture and a plurality of pictures including a P-picture And a setting unit for setting to be configured.
[0016]
In the present invention, when it is detected that a plurality of frames are continuously changed within the GOP time, it is determined that the scene change is continuous, and an encoded signal in which all GOPs are composed of I pictures is output. If no scene change is detected, or if a scene change is not detected in multiple frames within the time of the GOP, it is determined that the scene change is discontinuous, and the GOP is determined by all three pictures or multiple pictures including I and P pictures. The configured encoded signal is output.
[0017]
In order to achieve the above object, the method of the present invention encodes a video signal using a correlation between an I picture that uses a spatial correlation in one frame and a previous I picture. A P picture that is a predicted frame to be encoded, and a B picture that is an interpolation frame to be encoded using the correlation between both directions of the I picture and the P picture. In a video encoding method for outputting an encoded signal in which a plurality of pictures up to the immediately preceding picture are grouped as GOPs, a first step of detecting a scene change of the video signal and outputting a scene change detection signal When, a second step of determining whether the scene change detection signal is continuous in a plurality of frames in the GOP time, the second step, Nchenji continuous and when it is determined in a third step of outputting the coded signal in which all the GOP is constituted by I-picture, the second step, when it is determined that a scene change discontinuously, GOP of three picture And a fourth step of outputting an encoded signal composed of a plurality of pictures including all or I pictures and P pictures.
[0018]
In the present invention, when it is detected that a plurality of frames are continuously changed within the GOP time, it is determined that the scene change is continuous, and an encoded signal in which all GOPs are composed of I pictures is output. If no scene change is detected, or if a scene change is not detected in multiple frames within the time of the GOP, it is determined that the scene change is discontinuous, and the GOP is determined by all three pictures or multiple pictures including I and P pictures. The configured encoded signal is output.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a video encoding apparatus according to the present invention. As shown in FIG. 1, the video encoding apparatus of the present invention is characterized in that it includes a scene change detection circuit 14, a scene change continuous detection circuit 15, a GOP structure setting circuit 16, and a compression method selection circuit 17. is there.
[0020]
In FIG. 1, for example, a video signal input from a monitoring camera (not shown) is input to a buffer memory / preprocess unit (BM) 1. Specifically, the buffer memory / pre-processing unit 1 is configured to include a pre-processing unit for performing filtering processing, down-sampling processing, and the like at the subsequent stage of the frame buffer. While being supplied to the detection circuit 14, a predetermined pre-process is performed and output to the subtracter 2 and the changeover terminal a of the changeover switch 18.
[0021]
The scene change detection circuit 14 determines a scene change of the input video signal by a known method to be described later, generates a high level detection signal when it is determined as a scene change, and generates a low level detection signal otherwise. Output to the detection circuit 15.
[0022]
As a method for detecting a scene change, for example, a method described in JP-A-8-56361 can be used. In this scene change detection method, the sum of squares (corresponding to the energy of the difference) of the difference (pixel value difference) between the currently input frame and the frame input immediately before is calculated, and the value is predetermined. If the value is greater than or equal to the value, it is determined that a scene change has occurred in the currently input frame, and a high-level detection signal is output. Otherwise, a low-level detection signal is output.
[0023]
When receiving a high level detection signal from the scene change detection circuit 14, the scene change continuous detection circuit 15 monitors the detection signal for a predetermined time. When a high level detection signal is received again within a predetermined time, it is determined that the scene change is continuous, and a high level continuous detection signal is output. On the other hand, the scene change continuous detection circuit 15 does not receive a high level detection signal from the scene change detection circuit 14 or outputs a high level detection signal again within a predetermined time even when it receives a high level detection signal. If not received, a low level continuous detection signal indicating that the sea change is not continuous is output.
[0024]
When the continuous detection signal input from the scene change continuous detection circuit 15 is at a high level, the GOP structure setting circuit 16 determines that the scene change is continuous and configures the GOP with only an I picture, and the continuous detection signal is at a low level. In this case, it is determined that the scene change is discontinuous, and the GOP is normally configured with three types of I picture, B picture, and P picture, but may be configured with only I picture and P picture.
[0025]
It is to be noted that whether or not a scene change is continued for a predetermined time is detected when a scene change is detected in only one frame in units of 1/30 seconds, and an input video to the buffer memory / preprocessing unit 1 This is because when a disturbance is added to the signal, a scene change may be erroneously detected, leading to a malfunction. This is because actual encoding requires several frames. Note that during a period in which it is detected whether or not scene changes are continued for a predetermined time, encoding is performed in the order of pictures such as IBBP or IPPPP.
[0026]
When the GOP structure is set by the GOP structure setting circuit 16, the compression method setting circuit 17 selects the compression method using the I picture, B picture, or P picture based on the GOP structure. When the structure of the I picture is set by the GOP structure setting circuit 16, the compression method selection circuit 17 selects the I picture as the compression method, switches the changeover switch 18 to the terminal a side, and sets the open / close switch 19 Turn off (open).
[0027]
On the other hand, the video signal output from the buffer memory / preprocessing unit 1 is input to the discrete cosine transform unit (DCT) 3 through the changeover switch 18 and converted into a frequency domain signal, and further the quantization unit (Q) 4 And quantized (bit converted). The output quantized signal of the quantizing unit 4 is input to a variable length coding unit (VLC) 5 and subjected to variable length coding.
[0028]
In this way, when an I picture is selected, the quantization unit 4 to the inverse quantization unit (IQ) 8, the inverse discrete cosine transform unit (IDCT) 9, the adder 10, the frame memory (FM) 11, the motion compensation prediction The feedback loop of the unit (MC) 13 and the subtracter 2 is not used, and an intra frame encoded signal that is encoded using the spatial correlation in one frame is output to the multiplexer (MUX) 6. The
[0029]
When the P picture is selected as the compression method by the compression method selection circuit 17, the input video signal is processed as follows. In this case, the changeover switch 18 is switched and connected to the terminal b side by the output signal of the compression method selection circuit 17, and the open / close switch 19 is turned on (closed). As a result, the output signal of the subtracter 2 is selected by the changeover switch 18 and input to the discrete cosine transform unit 3, and the output signal of the motion compensation prediction unit 13 is fed back to the adder 10 through the open / close switch 19.
[0030]
The subtracter 2 receives the video signal of the nth frame once stored in the buffer memory / preprocessing unit 1 for each macro block. A macroblock is a unit used for motion compensation prediction, and is composed of 16 × 16 pixels or 16 × 8 pixels in the MPEG2 compression method. The subtractor 2 is stored in a frame memory (FM) 11 to be described later, and the locally decoded data of the ni (i ≠ 0) frame that has been motion compensation predicted by the motion compensation prediction unit (MC) 13 and the above-mentioned The difference from the video signal of the nth frame from the buffer memory / pre-processing unit 1 is obtained. This eliminates temporal redundancy.
[0031]
The output signal of the subtracter 2 is input to the discrete cosine transform unit (DCT) 3 through the changeover switch 18 to be converted into a frequency domain signal, and further input to the quantization unit (Q) 4 for quantization (bit conversion). ) The quantized signal output from the quantizing unit 4 is input to the variable length encoding unit (VLC) 5 to be variable length encoded, and is also input to the inverse quantizing unit (IQ) 8 to perform inverse quantization. Is done.
[0032]
The output signal of the inverse quantization unit 8 is input to the inverse discrete cosine transform unit (IDCT) 9 and returned to the state before the discrete cosine transform. The output signal of the inverse discrete cosine transform unit 9 is input to the adder 10, where it is added to the locally decoded data that is the output of the motion compensation prediction unit 13 input through the open / close switch 19, and then the addition result Is supplied to the frame memory 11.
[0033]
The motion vector detection unit 12 detects a motion vector based on the output video signal of the buffer memory / preprocessing unit 1 and the output signal of the frame memory 11 and supplies the detected motion vector to the motion compensation prediction unit 13. The motion compensation prediction unit 13 supplies the locally decoded data of the n−i (i ≠ 0) frame subjected to the motion compensation prediction to the subtracter 2 and variable information such as a motion vector used for encoding. This is supplied to the long encoding unit 5. The incidental information supplied from the motion compensation prediction unit 13 and the quantized signal supplied from the quantization unit 4 are input to the variable length coding unit 5 and subjected to variable length coding.
[0034]
When the compression method selection circuit 17 selects a B picture as the compression method, the input video signal is processed as follows. In this case, the changeover switch 18 is switched and connected to the terminal b side by the output signal of the compression method selection circuit 17, and the open / close switch 19 is turned on (closed). As a result, the output signal of the subtracter 2 is selected by the changeover switch 18 and input to the discrete cosine transform unit 3, and the output signal of the motion compensation prediction unit 13 is fed back to the adder 10 through the open / close switch 19.
[0035]
The subtracter 2 receives the video signal of the nth frame once stored in the buffer memory / preprocess unit 1 for each macro block. The subtractor 2 is stored in a frame memory (FM) 11 to be described later, and is locally decoded for the (n−i) th or n + j (i, j ≠ 0) frames that have been motion compensation predicted by the motion compensation prediction unit (MC) 13. The difference between the received data and the video signal of the nth frame from the buffer memory / preprocessing unit 1 is obtained.
[0036]
The output signal of the subtracter 2 is input to the discrete cosine transform unit (DCT) 3 and converted to a frequency domain signal, and further input to the quantization unit (Q) 4 and quantized (bit converted). The output quantized signal of the quantizing unit 4 is input to the variable length naming unit (VLC) 5 and also input to the inverse quantizing unit (IQ) 8 to be dequantized. The output signal of the inverse quantization unit 8 is input to the inverse discrete cosine transform unit (IDCT) 9 and returned to the state before the discrete cosine transform.
[0037]
The output of the inverse discrete cosine transform unit 9 is input to the adder 10, where it is added to the locally decoded data that is the output of the motion compensation prediction unit 13 input through the open / close switch 19, and the addition result is It is supplied to the frame memory 11.
[0038]
The motion vector detection unit 12 detects a motion vector based on the output video signal of the buffer memory / preprocessing unit 1 and the output of the frame memory 11 and supplies the motion vector to the motion compensation prediction unit 13.
[0039]
The motion compensation prediction unit 13 supplies incidental information such as a motion vector used for encoding to the variable length encoding unit 5. The incidental information supplied from the motion compensation prediction unit 13 and the quantized signal supplied from the quantization unit 4 are input to the variable length coding unit 5 and subjected to variable length coding.
[0040]
A signal that is variable-length-coded in the above-described procedure for each of the I picture, P picture, and B picture is input to the multiplexer (MUX) 6. The multiplexer 6 multiplexes the output signal of the variable length encoding unit 5 and the output of a speech encoded signal (not shown) and outputs the encoded data.
[0041]
Further, the code amount control unit (CC) 7 is necessary for the quantization in the quantization unit 4 so that the data amount per unit time of the encoded data falls within an appropriate range based on the information from the multiplexer 6. Set the coefficient.
[0042]
As described above, according to the present embodiment, when it is detected that a scene change has continued for a plurality of frames within the time of the GOP, it is determined that the scene change is continuous (that is, a scene where different images are switched). Te, continuous period of the scene change, since the outputs a coded signal composed of I-picture that does not use the correlation between before all GOP pictures, a plurality of frames in a plurality of images in the GOP time It can be applied to a monitoring VDR that records a signal switched in a short time, such as when a scene change is not detected, or when a scene change is no longer detected in multiple frames within the GOP time. GOP is composed of all three pictures or multiple pictures including I picture and P picture. Since it is possible to output a coded signal based on conventional MPEG scheme, when applied to a monitoring camera, both can record a moving image captured by the monitoring camera, more efficiently than the conventional JPEG method ( High compression ratio can be recorded.
[0043]
Next, a video encoding method according to the present invention will be described. FIG. 2 shows a flowchart of an embodiment of a video encoding method according to the present invention. In this embodiment, in a video encoding method for compressing a video signal by the MPEG method, first, a scene change of the input video signal is detected and a scene change detection signal is output (step S1). Subsequently, it is determined whether or not the scene change detection signal is continued within a predetermined time (step S2).
[0044]
If it is determined in step S2 that the scene change is continuous, an encoded signal in which all GOPs are composed of I pictures is output (step S3). On the other hand, when it is determined in step S2 that the scene change is not continuous, the GOP is composed of a plurality of pictures including all three pictures (I, P, and B pictures) or a plurality of pictures including I pictures and P pictures. The encoded signal is output (step S4).
[0045]
Thereby, according to the encoding method of this Embodiment, it can apply to the monitoring VDR similarly to the encoding apparatus of said Embodiment .
[0046]
【The invention's effect】
As described above, according to the present invention, when it is detected that a scene change is continued for a plurality of frames within the GOP time, it is determined that the scene change is a switching scene, and the continuous period of the scene change is When an encoded signal in which all GOPs are composed of I pictures is output and no scene change is detected , or when no scene change is detected in a plurality of frames within the GOP time, it is determined that the scene change is discontinuous and the GOP Since the encoded signal composed of all the three pictures or a plurality of pictures including the I picture and the P picture is output , it is more efficient than the JPEG encoding only, and the coding is free of visual defects. A video signal or continuous motion in which multiple images are switched in a short time such as multiple frames within the time in the GOP. The image signal can be recorded in coded encoded signal using an MPEG method.
[0047]
As a result, according to the present invention, the present invention can be applied to a monitoring VDR that records a video signal obtained by switching a plurality of images in a short time such as a plurality of frames within a time in a GOP. Can record moving image signals on a recording medium at a higher efficiency than the JPEG method adopted by JPEG, and it is not necessary to use two types of compression methods, the JPEG method and the MPEG method, in one recording / reproducing apparatus. Therefore, the apparatus can be configured at low cost, and compatibility of recorded data is possible, which is particularly suitable when applied to a monitoring VDR.
[Brief description of the drawings]
FIG. 1 is a block diagram of an embodiment of a device of the present invention.
FIG. 2 is a flowchart of an embodiment of the method of the present invention.
FIG. 3 is a system configuration diagram of an example of a monitoring system.
FIG. 4 is a block diagram of an example of an image signal recording / reproducing apparatus for monitoring.
[Explanation of symbols]
1 Frame buffer / preprocess part (BM)
2 Subtractor 3 Discrete cosine transform unit (DCT)
4 Quantizer (Q)
5 Variable length coding unit (VLC)
6 Multiplexer (MUX)
7 Code amount control unit (CC)
8 Inverse quantization unit (IQ)
9 Inverse discrete cosine transform (IDCT)
10 Adder 11 Frame memory (FM)
12 motion vector detection unit 13 motion compensation prediction unit (MC)
14 Scene change detection circuit 15 Scene change continuous detection circuit 16 GOP structure setting circuit 17 Compression method selection circuit 18 Changeover switch 19 Open / close switch

Claims (2)

An I picture that encodes a video signal using a spatial correlation within one frame, a P picture that is a prediction frame that encodes a video signal using a correlation with the previous I picture, and the I picture A plurality of pictures from the I picture to the picture immediately before the next I picture, encoded into any of the three pictures of the B picture, which is an interpolated frame that is encoded using the bidirectional correlation with the P picture In a video encoding device that outputs an encoded signal in which GOP is a group unit GOP ,
A scene change detector for detecting a scene change of the video signal;
When the scene change detection signal by the scene change detection unit continues for a plurality of frames within the time of the GOP, it is determined that the scene change is continuous, and when the detection signal indicating the scene change is not input, or a plurality of within the time of the GOP A scene change continuation determining unit that determines that the scene change is discontinuous when a detection signal indicating the scene change is not input in the frame ;
When the scene change continuation determination unit determines the continuity of the scene change, all the GOPs are composed of the I pictures, and when the scene change continuation determination unit determines the non-continuation of the scene change, the GOP And a setting unit configured to configure the plurality of pictures including the I picture and the P picture. An I picture that encodes a video signal using a spatial correlation within one frame, a P picture that is a prediction frame that encodes a video signal using a correlation with the previous I picture, and the I picture A plurality of pictures from the I picture to the picture immediately before the next I picture, encoded into any of the three pictures of the B picture, which is an interpolated frame that is encoded using the bidirectional correlation with the P picture In a video encoding method for outputting an encoded signal in which GOP is a group unit GOP ,
A first step of detecting a scene change of the video signal and outputting a scene change detection signal;
A second step of determining whether or not the scene change detection signal continues in a plurality of frames within the time of the GOP ;
When it is determined in the second step that the scene change is continuous, a third step of outputting an encoded signal in which the GOP is entirely composed of I pictures;
When it is determined in the second step that the scene change is not continuous, the GOP outputs an encoded signal composed of all the three pictures or the plurality of pictures including the I picture and the P picture. A video encoding method comprising the steps of:

Images