JP4293151B2 - Video encoding method and video recording apparatus - Google Patents

Description

  The present invention relates to a video encoding method and a video recording apparatus, and more particularly to a video encoding method suitable for generating an intermittent video signal by using inter-frame compression for a standard television signal related to a video and its intermittent video. The present invention relates to a video recording apparatus for recording a signal.

  In supermarkets, convenience stores, pachinko parlors, etc., camera surveillance systems are used to prevent shoplifting crimes and preserve evidence. In the past, this monitoring system used a time-lapse VTR, which is a VTR for long-term recording monitoring. However, since magnetic tape is used as a recording medium, problems such as clogging of the recording / reproducing head and tape entrainment occur. For this reason, in recent years, video disk recorders (hereinafter referred to as VDR) that record image signals on a hard disk or an optical disk are increasingly used in surveillance systems.

  Also in this VDR, as in the previous time-lapse VTR, a method of intermittently recording a moving image is employed in order to extend the recording time. 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. In such a VDR, for every n frames (n is a natural number of 2 or more) of the input video signal, the first frame is encoded into an intra frame, and (n-1) frames after the first frame are encoded. Is encoded into a P picture, which is a prediction frame using a correlation with an intra frame, and the encoded frames are thinned out at a predetermined rate to obtain one intra frame and m (m is equal to or less than n−1). A video encoding method for outputting a code sequence including one intra frame and m prediction frames as a (natural number) prediction frame has been previously proposed by the present applicant (see, for example, Patent Document 1).

  According to this conventional video coding method, P-pictures are thinned out, and image data compressed in a frame by JPEG (Joint Photographic Experts Group) method or in-frame compression of only an I picture that is an MPEG2 method intra frame. Compared to recording image data, it is possible to record moving image signals with high image quality and high compression at a low frame rate.

JP 2003-333526 A

  However, in the conventional video encoding method described in Patent Document 1, the compression rate and the image quality change depending on the type of image such as a moving image or a still image. FIG. 8 is a schematic diagram showing the relationship between the conventional frame and the data amount. FIG. 8A shows an example in which a still image is compressed, and the data amount of the P picture is small. In general, in MPEG compression, code amount control is often performed so that the amount of data becomes almost constant in GOP (Group Of Picture) units.

  According to FIG. 8A, since the data amount of the P picture is small, the data amount of the entire GOP is small. Therefore, when code amount control is performed so that the data amount is constant in GOP units, the entire data amount increases as shown in FIG. 8B, and the data amount of the I picture increases. If the P picture of this image is thinned out, the result is as shown in FIG. 8C, and the effect of reducing the data amount by thinning out is small, and the still image has higher image quality than necessary.

  Next, FIG. 8D shows an example in which a moving image is compressed. Since the data amount of the P picture is large, the data amount of the entire GOP is large. When code amount control is performed on this image so that the data amount becomes constant in GOP units, the data amount as a whole decreases as shown in FIG. 8E, and the data amount of the I picture decreases. When the P picture of this image is thinned out, the result is as shown in FIG. 8F, and although the data amount is reduced, the data amount of the I picture is small, so the image quality of the moving image is degraded.

  For monitoring purposes, there is no subject movement in normal cases where there is no abnormality, and there are many cases where there is movement of the subject when an abnormality occurs and you want to see an image. It has been. Therefore, the conventional video coding method in which the image quality and the compression rate change, and particularly the moving image quality deteriorates, has a problem that it does not sufficiently meet the above-described demands for monitoring applications.

  The present invention has been made in view of the above points. When an intermittent video signal is generated using inter-frame compression of an input video signal, there is little variation in image quality between a still image and a moving image. An object of the present invention is to provide a video encoding method that is particularly suitable for monitoring applications and has a small amount of code data and high image quality for moving images, and a video recording apparatus that records the intermittent moving image signal.

  In order to achieve the above object, the video encoding method of the present invention includes an intra frame obtained by intra-frame compression encoding an input video signal, and a code of any prediction frame obtained by predictive encoding using a correlation with the intra frame. A video encoding method for selectively performing encoding, wherein n consecutive frames of an input video signal (n is a natural number of 2 or more) are used as encoding group units, and the first frame is determined for each encoding group unit. The first step of encoding into an intra frame while performing the code amount control to bring the data amount close to a desired target value, and the remaining (n−1) excluding the first frame for each encoding group unit A second step of encoding a frame into a prediction frame without performing code amount control and a prediction frame encoded in the second step are thinned out at a preset ratio. A third step, and a fourth step of outputting a code string composed of the intra frame obtained in the first step and the remaining prediction frames thinned out in the third step in a predetermined format. It is characterized by that.

  In the present invention, when outputting a code string composed of an intra frame and a prediction frame obtained by thinning out (n-1) prediction frames, encoding is performed so that the data amount of the intra frame approaches the target value. As compared with the conventional case where encoding is performed without considering the data amount of the intra frame, it is possible to generate an intermittent moving image signal having a good image quality with a small data amount.

  In order to achieve the above object, the video recording apparatus according to the present invention includes an intra frame obtained by intra-frame compression coding an input video signal and a prediction frame obtained by predictive coding using a correlation with the intra frame. A video recording apparatus for recording a code string obtained by selectively performing encoding on a recording medium, wherein n consecutive frames of an input video signal (n is a natural number of 2 or more) are encoded group units, For each encoding group unit, the first encoding unit that encodes the first frame into an intra frame while performing code amount control to bring the data amount close to a desired target value, and for each encoding group unit, The remaining (n-1) frames excluding the first frame are encoded by a second encoding unit that encodes the prediction frame without performing the code amount control, and the second encoding unit A code sequence comprising a frame decimation unit that decimates the predicted frame obtained at a preset ratio, an intra frame obtained by the first encoding unit, and a remaining prediction frame decimation by the frame decimation unit And recording means for recording on a recording medium in the format described above.

  In the present invention, when a code string composed of an intra frame and a prediction frame obtained by thinning out (n-1) prediction frames is recorded on a recording medium, encoding is performed so that the data amount of the intra frame approaches a target value. By doing so, it is possible to record a code string for a long time with high image quality on a recording medium of the same capacity as compared to the case of recording a code string encoded without considering the amount of data of an intra frame as in the past. .

  According to the present invention, encoding is performed so that the data amount of an intra frame approaches a target value, code amount control for each encoding group is not performed, and code amount control is not performed for a prediction frame, As a large amount of data, a part of the prediction frame is thinned and output or recorded, so the intraframe of the moving image has the same code amount as the intraframe of the still image, so the image quality does not deteriorate and the movement is large. Since a predicted frame of an image has a large amount of data, a video signal with sufficient image quality can be output or recorded with respect to motion, and is particularly suitable for application to monitoring.

  Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an image compression unit to which an embodiment of a video encoding method according to the present invention is applied. In the figure, the image format conversion unit 101 in the image compression unit 1 converts the supplied original image data into the luminance signal Y from the RGB format of the three primary color signals and the color difference format other than the sampling ratio of 4: 2: 0. The image data is converted into 4: 2: 0 YCbCr format image data consisting of the color difference signals Cb and Cr, and only Cb and Cr are level-shifted in the positive direction by 0x80 (hexadecimal number), for example, and output.

  The two-dimensional DCT processing unit 102 separately supplies each Y / Cb / Cr image data of, for example, 4: 2: 0 YCbCr format supplied from the image format conversion unit 101 in units of blocks. (Cosine transform) processing and output each DCT coefficient. The buffer 103 temporarily stores at least one screen of DCT coefficients for each DCT coefficient of Y / Cb / Cr image data supplied from the two-dimensional DCT processing unit 102.

  The quantization processing unit 104 quantizes each DCT coefficient read from the buffer unit 103 and outputs the compressed data to the entropy coding unit 105. The entropy encoding unit 105 entropy-encodes each compressed data of the Y, Cb, and Cr image data supplied from the quantization processing unit 104, and outputs the further compressed code data to the output buffer unit 106. To do. The output buffer unit 106 temporarily stores each code data supplied from the entropy encoding unit 105.

  Furthermore, the code amount determination unit 107 sets a final target code amount, and based on the code amount of each code data supplied from the entropy encoding unit 105, the target code amount for the luminance signal data (Y data). To decide. The quantization table calculation unit 108, based on the target code amount set and determined by the code amount determination unit 107 and the quantization table set in the quantization table setting unit 109 described later, the DCT coefficient of the luminance signal data. A quantization table used in the quantization process is generated and set in the quantization processing unit 104.

  The quantization table storage unit 110 stores a default quantization table that is used when the DCT coefficients of at least Y / Cb / Cr image data are first quantized. Based on the control of the quantization table setting unit 109, the read control unit 111 performs quantization in the DCT coefficient of each image data of Y / Cb / Cr stored in the buffer unit 103 by the quantization processing unit 104. Control is performed to select and read out DCT coefficients to be used for processing.

  Next, an embodiment of a video encoding apparatus according to the present invention will be described. FIG. 2 shows a block diagram of an embodiment of a video encoding apparatus according to the present invention. In the figure, a VDR 20 is a recording / reproducing apparatus (video disc recorder) having not only an intermittent moving picture recording function as an embodiment of the video encoding apparatus according to the present invention but also a reproducing function. 1, the image compression unit 22, the audio compression unit 23, the thinning processing unit 24, the disk device 25, the image expansion unit 26, the audio expansion unit 27, the D / A conversion unit 28, the CPU ( A central processing unit) 2A, an operation unit 2B, and a display unit 2C. These units 21 to 28, 2A, 2B, and 2C are configured to be centrally controlled by the CPU 2A via a bidirectional bus 29. ing.

  The image compression unit 22 and the audio compression unit 23 are configured by a general-purpose MPEG compression integrated circuit (IC) that performs compression encoding on a standard television signal in accordance with a known MPEG2 system. The operation unit 2B is operated by a user and inputs designation of a recording mode or a reproduction mode, the number of frames to be recorded, and the like. The CPU 2A controls necessary blocks among the above-described units in accordance with an input instruction from the operation unit 2B. The display unit 2C displays the status of the VDR 20.

  Next, the operation of the embodiment of FIG. 2 will be described with reference to the time chart of FIG. First, the operation during recording will be described. For example, an NTSC standard television signal obtained by imaging a subject such as a monitoring target with a camera (not shown) is supplied to the A / D converter 21 in the VDR 20, and the image signal is digitized only in necessary frames. After that, the audio signal is supplied to the image compression unit 22, and the audio signal is digitized and then supplied to the audio compression unit 23.

  FIG. 3 (a) schematically shows a standard television signal input to the VDR 20 in units of one frame. Here, television signals (image signals and audio signals) for four frames of V1, V2, V3, and V4 are shown. In addition, the degree of thinning by the thinning processing unit 24 described later is set to 1/3 (2 frames are thinned out of 3 frames).

  The image compression unit 22 compresses and encodes the digitized image signal input from the A / D conversion unit 21 in accordance with the MPEG2-PS (Program Stream) system with the configuration of the block diagram of FIG. However, in this embodiment, for each predetermined consecutive n frames (n is a natural number of 2 or more) of the input standard television signal related to a moving image, the top one frame is encoded using one spatial correlation. The remaining (n-1) frames are predicted frames obtained by encoding using the correlation with the I picture, that is, P pictures. In the example of FIG. 3, the above n is 6. Therefore, the image compression unit 22 encodes the first frame V1 shown in FIG. 3A into an I picture. In encoding an I picture, the code amount is controlled so as to approach the data amount of a desired target value.

  The operation of encoding the I picture by the image compression unit 22 will be described in detail with reference to the block diagram of FIG. 1 and the flowcharts of FIGS. 4 and 5. First, the image compression unit 22 is supplied to the image compression unit 22. The RGB format image data is converted from the RGB format to the YCbCr format by the image format conversion unit 101 in FIG. This format conversion can be easily converted by a known matrix conversion formula, and detailed description of the matrix conversion is omitted.

  Next, with respect to the converted YCbCr data, pixels are not thinned out for Y (luminance) data, and every other pixel is thinned out in the horizontal and vertical directions for Cb and Cr (both color difference) data. Thus, YCbCr data with a sampling ratio of 4: 2: 0 is generated. Subsequently, the generated Cb and Cr data are level-shifted by 0x80 in the plus direction and output in order to facilitate the two-dimensional DCT processing in the two-dimensional DCT processing unit 102 shown in FIG. Step S201 in FIG.

  In the generation of the YCbCr data with the sampling ratio 4: 2: 0, as shown in FIG. 6, the Y data has four block data Y0, Y1, Y2 and Y3 are generated. Therefore, the Y / Cb / Cr block data output from the image format conversion unit 101 will be described below as Y0, Y1, Y2, and Y3, and Cb and Cr.

  Next, the two-dimensional DCT processing unit 102 performs two-dimensional DCT processing on a block-by-block basis for each screen data for Cb and Cr output from the image format conversion unit 101 and Y0 to Y3, respectively. The DCT coefficient is stored in the buffer unit 103 shown in FIG. 1 (step S202 in FIG. 4). At this time, each DCT coefficient stored in the buffer unit 103 is stored in a state in which it can be read by the read control unit 111.

  Next, the quantization table setting unit 109 in FIG. 1 reads the default quantization table for Cb data stored in advance in the quantization table storage unit 110 and sets it in the quantization processing unit 104 (step in FIG. 4). S203). Subsequently, the quantization table setting unit 109 controls the read control unit 111 to continuously read out the DCT coefficients of the Cb data from the buffer unit 103. As a result, the read control unit 111 controls the address of the buffer unit 103 to read out the DCT coefficient of the Cb data from the buffer unit 103 and supply the DCT coefficient to the quantization processing unit 104. Here, the default quantization table for Cb data is used. Is quantized based on (step S204 in FIG. 4).

  Next, the Cb quantized data output from the quantization processing unit 104 is Huffman encoded by the entropy encoding unit 105 in FIG. 1 and further compressed (step S205 in FIG. 4). The Cb code data Huffman-encoded by the entropy encoding unit 105 is byte-stuffed as necessary and temporarily stored in the output buffer unit 106, and the data size SIZEbcb of the Cb code data is the code amount determination unit. 107 (steps S206 and S207 in FIG. 4).

  Next, the quantization table setting unit 109 in FIG. 1 reads the default quantization table for Cr data stored in advance in the quantization table storage unit 110 and sets it in the quantization processing unit 104 (step in FIG. 4). S208). In this case, the default quantization table for Cr data may be the same as the default quantization table for Cb data described above. Since the subsequent processing relating to the Cr data (steps S209 to S212 in FIG. 4) is the same as the processing relating to the Cb data (steps S204 to S207 in FIG. 4), description thereof is omitted. However, the data size of the Cr code data stored in the code amount determination unit 107 is SIZEbcr.

  As described above, the code data after compression of the color difference data (Cb and Cr) for one screen is temporarily stored in the output buffer unit 106, and each code amount SIZEbcb and SIZEbcr and the overall target code amount TARGET for one screen. Are set and stored in the code amount determination unit 107.

  Next, the image compression unit 1 (22 in FIG. 2) in FIG. 1 performs the data compression process of Y data according to the flowchart in FIG. 5 following the process in step S212 in FIG. First, the code amount determination unit 107 in FIG. 1 calculates the target code amount TARGETy of Y data from the overall target code amount TARGET for one screen and the code amounts SIZEbbc and SIZEbcr for one screen of Cb and Cr as follows. It is obtained by a calculation formula and stored (step S213 in FIG. 5).

TARGETy = TARGET- (SIZEbbcb + SIZEbcr)
Next, for the Y data, the luminance data divided into Y0, Y1, Y2, and Y3 shown in FIG. 6 are subjected to DCT coefficient quantization and entropy coding in the order of Y0, Y1, Y2, and Y3. Execute each process. First, the quantization table setting unit 109 reads the default quantization table (Q0y) for Y data stored in advance in the quantization table storage unit 110 and sets it in the quantization processing unit 104 (step S214 in FIG. 5). ).

  Subsequently, the quantization table setting unit 109 controls the read control unit 111 so as to continuously read the DCT coefficients of the Y0 data from the buffer unit 103, whereby the read control unit 111 controls the address of the buffer unit 103. Thus, the DCT coefficient of the Y0 data is read from the buffer unit 103 and supplied to the quantization processing unit 104, where it is quantized (step S215 in FIG. 5). Subsequently, the DCT coefficient of the quantized Y0 data is Huffman encoded by the entropy encoding unit 105, and the data size SIZE0y of the obtained code data is supplied to the code amount determining unit 107 (step S216 in FIG. 5). ).

  Next, the code amount determination unit 107 compares the data size SIZE0y of the Y0 code data with the target code amount TARGETy of the Y data, and when it is determined that, for example, SIZE0y is within ± 5% of TARGETy, Control is performed so as to stop the subsequent processing of Y1, Y2, and Y3 (step S217 Yes in FIG. 5). On the other hand, when the code amount determination unit 107 determines that SIZE0y is not within ± 5% of TARGETy (No in step S217), the determination result is supplied to the quantization table calculation unit 108, and the quantization table calculation unit 108 The quantization table Q1y for Y1 is calculated by the following calculation formula (step S218 in FIG. 5).

Q1y = ((1/4) × TARGETy / SIZE0y) × Q0
Next, the quantization table setting unit 109 sets the Y1 quantization table Q1y calculated by the quantization table calculation unit 108 in the quantization processing unit 104 (step S219 in FIG. 5). Subsequently, as in steps S215 to S217, the quantization processing unit 104 performs the DCT coefficient quantization processing on the Y1 data (step S220 in FIG. 5), and the entropy encoding unit 105 encodes the Y1 quantization data. (Step S221 in FIG. 5), the code amount determination unit 107 determines whether the data size SIZE1y of the Y1 code data is within ± 5% of the target code amount TARGETy of the Y data (Step in FIG. 5) S222) is performed sequentially.

  When SIZE1y is within ± 5% of the target code amount TARGETy, the subsequent processing is stopped. When SIZE1y is larger than ± 5%, the quantization table calculation unit 108 performs the quantization for Y2 according to the following equation: The table Q2y is calculated (step S223 in FIG. 5).

Q2y = ((1/4) × TARGETy / SIZE1y) × Q1y
Next, the quantization table setting unit 109 sets the Y2 quantization table Q2y calculated by the quantization table calculation unit 108 in the quantization processing unit 104 (step S224 in FIG. 5). Subsequently, similarly to steps S215 to S217, the quantization process by the quantization processing unit 104 of the DCT coefficient of the Y2 data (step S225 in FIG. 5), and the encoding process of the Y2 quantized data by the entropy coding unit 105 (Step S226 in FIG. 5), the code amount determination unit 107 determines whether the data size SIZE2y of the Y2 code data is within ± 5% of the target code amount TARGETy of the Y data (Step in FIG. 5) S227) is performed sequentially.

  When SIZE2y is within ± 5% of the target code amount TARGETy, the subsequent processing is stopped. When SIZE2y is greater than ± 5%, the quantization table calculation unit 108 performs quantization for Y3 according to the following equation: The table Q3y is calculated (step S228 in FIG. 5).

Q3y = ((1/4) × TARGETy / SIZE2y) × Q2y
Next, the quantization table setting unit 109 sets the Y3 quantization table Q3y calculated by the quantization table calculation unit 108 in the quantization processing unit 104 (step S229 in FIG. 5). Subsequently, similarly to steps S215 and S216, the quantization processing unit 104 performs the DCT coefficient quantization processing on the Y3 data (step S230 in FIG. 5), and the entropy encoding unit 105 encodes the Y3 quantization data. (Step S231 in FIG. 5) is sequentially performed.

  Then, the quantization table calculation unit 108 uses the data size SIZE3y of the Y3 code data determined by the code amount determination unit 107 to calculate the final quantization table Qy for the entire Y data by the following calculation formula: (Step S232 in FIG. 5).

Qy = ((1/4) × TARGETy / SIZE3y) × Q3y
Subsequently, the final quantization table Qy for the entire Y data is set in the quantization processing unit 104 by the quantization table setting unit 109 (step S233 in FIG. 5). For all DCT coefficients, quantization processing by the quantization processing unit 104 (step S234 in FIG. 5) and quantization processing of quantized data of Y data by the entropy coding unit 105 (step S235 in FIG. 5) are sequentially performed. Done.

  If it is determined that SIZE0y, SIZE1y, or SIZE2y is within ± 5% of TARGETy and branching is transferred from step S217, S222, or S227 to step S234, the quantization table Q1y, Q2y or Q3y is used as Qy.

  Subsequent to the processing in step S235 described above, the final Y code data Huffman-encoded by the entropy encoding unit 105 is byte-stuffed as necessary and supplied to the output buffer unit 106, and is already temporarily stored. The compressed Cb and Cr code data are output as compressed data (step S236 in FIG. 5). In this way, by processing the I picture, it is possible to output code data having a code amount that is the closest to the target code amount.

  Next, the image compression unit 22 in FIG. 2 encodes the frames V2, V3, V4,..., V (n−1) after the second frame shown in FIG. The prediction frame to be encoded using the correlation with the obtained I picture, that is, the P picture is encoded. In coding a P picture, code amount control is not performed, and the same processing is always performed and coding is performed by a known method. As a result, the size of the P picture is increased in a moving image with intense motion.

  FIG. 7A shows an example of the relationship between the data amount obtained by encoding a still image and the frame, and FIG. 7C shows an example of the relationship between the data amount obtained by encoding a moving image and the frame. The above encoding is performed every n frames of the input standard television signal (n = 6 in the example of FIG. 7). Since encoding is performed in this way, the entire data stream is encoded not as CBR (constant bit rate) but as VBR (variable bit rate).

  Each picture is composed of a plurality of video packets (video PES (Packetized Elementary Stream)). 2 compresses and encodes the digitized audio signal input from the A / D converter 21 and outputs an audio pack. This audio pack is put into the data stream together with the video packet output from the image compression unit 22.

  Each audio pack and video packet includes PTS (Presentation Time Stamp) data representing the presentation time. In summary, the output signals of the image compression unit 22 and the audio compression unit 23 are as shown in FIG. In FIG. PES is a video PES, A. PES schematically represents an audio PES. However, in FIG. 3B, only the audio PES is an output packet of the audio compression unit 23, and the other is an output packet of the image compression unit 22. Further, pa1 to pa4 are padding packets for making the entire bit rate constant in the case of a fixed bit rate.

  In FIG. 2, the compression-encoded image data (PS) output from the image compression unit 22 is supplied to the thinning processing unit 24, where every three consecutive frames including the first I frame I picture. Two frames of video packets (V.PES) excluding the first frame and padding packets of all frames are thinned out, and then supplied to the disk device 25. On the other hand, at the same time, the compression-coded audio data (audio PES) output from the audio compression unit 23 is supplied to the disk device 25 without being thinned out.

  Thereby, as schematically shown in FIG. 3 (c), the disk device 25 is composed of an I picture of the first frame, a video PES of P picture of one frame per three frames, and all audio PESs. Discrete data is input and recorded on a hard disk or an optical disk. That is, the disk device 25 records, on a hard disk or an optical disk, a packet in which inter-frame compression of an intermittent moving image in which an I picture and a P picture are synthesized in time series is performed. FIG. 7B shows an example of data obtained by thinning out a still image and shows an example of thinning out a moving image, which shows the relationship between a frame and a data amount recorded on a hard disk or an optical disk.

  Next, the operation of the reproduction system of FIG. 2 will be described. The CPU 2A controls the disk device 25 to reproduce the data recorded on the hard disk or the optical disk. Among the obtained reproduction data, the video PES of I picture or P picture is supplied to the image expansion unit 26, and the audio PES is audio. Supply to the extension unit 27.

  Each of the expansion units 26 and 27 outputs the access unit when an internal reference synchronization signal STC (System Time Clock) reaches the PTS in the input reproduction data. Thereby, the image decompression unit 26 decodes, for example, the I picture and the P picture of FIG. 3C, decodes the first frame V1 and the subsequent V4 shown in FIG. / A converter 28 is supplied.

  The D / A converter 28 converts the input decoded frames V1 and V4 into analog image signals, and interpolates the thinned second and third frames V2 and V3 with the analog image signals of the first frame V1. Then, interpolation processing such as interpolating the fifth and sixth frames with the analog image signal of the fourth frame V4 is performed. As shown in FIG. 3E, the signal is an NTSC standard television signal of 30 frames per second. Thus, as a content image, an intermittent moving image signal of 10 frames per second is output.

  On the other hand, since the audio PES has not been thinned out, the audio data obtained by being decompressed by the speech decompression unit 27 is converted into an uninterrupted analog speech signal by the D / A conversion unit 28 as shown in FIG. Is converted to output.

  Here, the video PES recorded on the hard disk or the optical disk reproduced by the disk device 25 is, for example, FIG. 7B in the case of data obtained by thinning out still images, and in the case of data in which moving pictures are thinned out, for example, FIG. d). As can be seen from FIGS. 7B and 7D, since the size of the I picture is fixed, variations in image quality between still images and moving images are reduced. In addition, since the size of a P picture is large in a moving image with a large amount of motion, the compression rate changes according to the image quality, but the bit rate increases as the motion increases.

  When monitoring is performed using video, there are many still images in the normal state, and there are many movements on the screen in the abnormal state. For this reason, when this embodiment is applied to a monitoring VDR, the amount of data decreases in a normal state, and the image quality becomes high in an abnormal state, which is particularly suitable for monitoring purposes.

  As described above, according to the present embodiment, the image compression unit 1 encodes the I picture by performing the code amount control so that the target data amount is obtained, and does not perform the GOP unit code amount control. Since the code amount control is not performed and a large motion image is recorded as a large amount of data, a part of the P picture is thinned out, so that the moving picture I picture also has the same code quantity as the still picture I picture. Since the image quality does not deteriorate and the P picture of an image with a large amount of motion has a large amount of data, the image can be recorded on the recording medium with a sufficient image quality regarding the motion.

  Note that the present invention is not limited to the above-described embodiment. For example, the image compression unit 22, the audio compression unit 23, the image expansion unit 26, and the audio expansion unit 27 are combined into one MPEG having an encoding / decoding function. A codec IC can be used. In the embodiment, the recording device uses a disk device 25 that records and reproduces a code string on a hard disk or an optical disk. However, a magnetic recording / reproducing device that records and reproduces a code string on a magnetic tape or a magnetic disk, and a semiconductor memory. Of course, the present invention can also be applied to a memory recording / reproducing apparatus that records and reproduces a code string.

  In the case of using a magnetic tape, there is a problem of the clogging of the head described above. However, according to the present invention, a code string having a larger capacity than that of the conventional recording can be recorded on a magnetic tape or magnetic disk having the same capacity as the conventional one. There is a feature that reproduction, that is, recording and reproduction for a long time is possible. The present invention also includes a computer program for realizing the operation of the image compression unit 1 (22 in FIG. 2) in FIG. 1 by a computer.

It is a block diagram of an image compression part to which an embodiment of the method of the present invention is applied. It is a block diagram of one embodiment of a video recording device of the present invention. 3 is a time chart for explaining operations of FIGS. 1 and 2. FIG. 3 is a flowchart (part 1) for explaining the operation of FIG. 1; FIG. FIG. 3 is a flowchart (part 2) for explaining the operation of FIG. 1; It is a schematic diagram of an example of the recorded image in this invention apparatus. It is a schematic diagram of the relationship between the flame | frame and data amount in one embodiment of this invention. It is a schematic diagram of the relationship between a conventional example frame and data amount.

Explanation of symbols

1, 22 Image compression unit 20 VDR
21 A / D conversion unit 23 Audio compression unit 24 Thinning-out processing unit 25 Disk device 26 Image expansion unit 27 Audio expansion unit 28 D / A conversion unit 2A Central processing unit (CPU)
2B Operation unit 2C Display unit 101 Image format conversion unit 102 Two-dimensional DCT processing unit 103 Buffer unit 104 Quantization processing unit 105 Entropy encoding unit 106 Output buffer unit 107 Code amount determination unit 108 Quantization table calculation unit 109 Quantization table setting Unit 110 Quantization table storage unit 111 Read control unit

Claims (2)

A video encoding method that selectively encodes one of an intra frame obtained by intra-frame compression encoding an input video signal and a prediction frame that is predictively encoded using a correlation with the intra frame,
Code amount control in which n frames (n is a natural number greater than or equal to 2) of the input video signal are set as a coding group unit, and the data amount of the first frame is brought close to a desired target value for each coding group unit. A first step of encoding into the intra frame while performing
A second step of encoding the remaining (n−1) frames excluding the first frame for each encoding group unit into the prediction frame without performing code amount control;
A third step of thinning out the prediction frame encoded in the second step at a preset rate;
And a fourth step of outputting a code string composed of the intra frame obtained in the first step and the remaining prediction frame thinned out in the third step in a predetermined format. A characteristic video encoding method. A code string obtained by selectively performing encoding of an intra frame obtained by intra-frame compression encoding of an input video signal and a prediction frame obtained by predictive encoding using a correlation with the intra frame is used as a recording medium. A video recording device for recording,
Code amount control in which n frames (n is a natural number greater than or equal to 2) of the input video signal are set as a coding group unit, and the data amount of the first frame is brought close to a desired target value for each coding group unit. First encoding means for encoding into the intra frame while performing
Second encoding means for encoding the remaining (n−1) frames excluding the first frame for each encoding group unit into the prediction frame without performing code amount control;
Frame thinning means for thinning out the prediction frame encoded by the second encoding means at a preset rate;
Recording means for recording a code string composed of the intra frame obtained by the first encoding means and the remaining prediction frames thinned by the frame thinning means on a recording medium in a predetermined format. A video recording apparatus characterized by that.

Images