JP2009017062A - High definition moving image recording and reproducing method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To efficiently compress a moving image when dividing a high definition moving image within a screen to compress the moving image. <P>SOLUTION: Screen division is performed on the moving image, and one screen obtained by screen division per frame is selected in order in accordance with a person area in the image, variance values per distributed image, difference values per distributed image, and frequency conversion coefficients of distributed images and is compressed by inter-image coding and is recorded together with divided screen position information. In the case of reproduction, only an expanded screen is updated by using recorded divided screen position information, and images expanded in the past are held and displayed with respect to the other divided screens. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present technology relates to a digital device such as a camera or a recorder capable of recording / reproducing moving images.

  As background art in this technical field, for example, there is JP-A 2000-165859 (Patent Document 1). In this publication, “[Problem] An image having different resolutions is compressed and decompressed by the same apparatus. [Solution means] The image data dividing means 1a, when the image data is high resolution image data, The image data compression means 1b performs compression processing on each partial image data or normal image data, and the specific information addition means 1c compresses the compressed image. If the data is partial image data, specific information for specifying the position in the high resolution image data is added, and the flag adding means 1d indicates that the partial image data has been divided. The flag detection unit 3a detects a flag from the compressed image data, and the specific information extraction unit 3b extracts the specific information from the image data when the flag is detected. The image data decompression unit 3c performs decompression processing on the image data, and the resynthesizing unit 3d resynthesizes the partial image data according to the specific information when the flag is detected, and performs high resolution. "Reproduce and output image data" (see summary).

JP 2000-165859 A

  In recent years, HD (High Definition) high-definition televisions, recorders, video cameras, and the like have become widespread. Furthermore, super-resolution moving images exceeding these image resolutions have attracted attention.

  On the other hand, as a compression technique for moving image encoding, H.264 is used. H.264 encoding has attracted attention. This can achieve about three times the compression compared to the conventional moving image coding, and will become a standard for future recorders and video cameras. However, in the case of a moving image having a resolution higher than HD, H.264 is used for transmission / reception and recording of moving images. An image compression capability of H.264 encoding or higher is required.

  As a method of efficiently compressing a super-resolution moving image using the current compression technique, for example, Patent Document 1 describes a method of dividing and compressing an image. However, there is no description about a picture configuration or a rate control method for efficiently compressing a divided image into a moving image. In order to compress the divided images in real time, H.264 is used. There arises a problem that an image compression capability higher than that of H.264 encoding is necessary, or a problem that the circuit scale increases due to the parallel processing of compression.

  Therefore, in the present invention, when a high-definition moving image is divided on the screen to perform moving image compression, efficient moving image compression is performed without increasing the image compression capability or circuit scale per frame. With the goal.

  The above object can be achieved by the invention described in the claims. For example, focusing on the correlation, person position, or motion vector in a high-definition moving image, the position of the divided screen selected for each frame when the moving image is encoded is controlled.

  High-definition moving images can be compressed and saved with high data quality with a small data size.

  Problems, configurations, and effects other than those described above will be clarified by the following description.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows an example of moving image division encoding. In this figure, one high-definition still image is divided into nine, and these are encoded by a moving image encoder. In this example, an H.264 encoder is used as the moving image encoder, and the image is divided into rectangular regions of the same size all in the image.

  Next, FIG. 2 shows a method of main moving image encoding.

When an image is divided into nine and they are encoded as moving images, the importance level is determined. The importance is an index for compressing to a higher image quality, and where the importance is high, the picture type is set to I picture or many bits are allocated at the rate control.
Regarding the determination of the importance, the following is used.
-Increase the importance of the split image at the center of the screen.
・ Identify a person through image recognition and increase the importance of that part.
-Increase the importance by specifying the area by the user.
・ Continuously analyze the motion of the image to increase the importance of the divided images that have moved.

  After determining the importance, the number of images to be encoded per second of the divided images is determined in descending order of the importance, and the position and order of the divided images to be selected for each frame are determined.

  Next, the picture type is determined.

In determining the picture type, the correlation of each divided image obtained by dividing the image is analyzed. The picture type is determined in descending order of importance for the divided image group rearranged in order of importance according to the correlation.
If the correlation with any other divided image is low, the divided image is set as an I picture.
When the correlation with other divided images is high, the first picture is an I picture, the second picture is a P picture, and the third and subsequent pictures are B pictures. At this time, if the reference picture for motion compensation of the P and B pictures is P, the first I picture is used. If it is B, it will be the first I picture and the next P picture.

The correlation analysis will be described. One or several of the following are used for the correlation analysis.
A variance value for each divided image is calculated, and those having similar values are assumed to have high correlation.
A person who has a small difference value by calculating a difference between divided images is assumed to have high correlation.
Assume that the DCT value is mapped for each coding target block in the divided image, and that the maps are similar to each other in the dispersed images are highly correlated.
Encoding is performed with the picture type determined as described above.

  Next, rate control during moving image encoding will be described. At the time of encoding, bit allocation of each picture is performed so that the entire compressed data amount of the image becomes a desired data amount. At this time, in the allocation of the bits of the I picture, P picture, and B picture, the bit allocation is increased as the divided image is closer to the head.

  Encoding is performed for each actual divided image. When encoding a divided image, quantization is performed in the upper, lower, left, and right coding unit blocks (MB: macroblock) adjacent to the screen edge in the divided image. The quantization value to be performed is set to the same value as the MB if adjacent MBs in the upper, lower, left, and right sides of the peripheral divided image have already been encoded. By doing so, distortion at the division boundary when decompressing still image information after compression is reduced. For MBs that are not adjacent, the quantization value for each MB is controlled so that the value does not change significantly from the quantization values of the neighboring divided images and the adjacent blocks.

  The concept is shown in FIG. In this figure, with respect to the encoding of the divided image, for the upper already encoded divided image A, the middle already encoded divided image B, and the upper already encoded divided image A, the encoding has been completed in the past frame. And When encoding, in this case, since the upper and lower and right divided images have already been encoded, the quantization values for the upper and lower and right end MBs in the divided image are already encoded. The value is the same as the quantized value of C MB.

  In the case of reproduction, only the expanded screen is updated using the recorded divided screen position information, and the other divided screens hold and display images expanded in the past.

  The first embodiment is effective in a moving image standard that can freely adjust the reference relationship between P and B pictures. However, in a compression standard such as MPEG2, the reference relationship between P and B is defined by the standard. The following method is effective when the present invention is used in such a standard.

First, the importance is determined as in the first embodiment. Next, the correlation is determined in the same manner as in the first embodiment, and groups having high correlations are configured.
Next, the number of times of encoding per second for each group is determined. The number of times of encoding per second for each group is assigned from a group including divided images with high importance. Finally, the position and picture type of the divided image to be selected for each frame in the group are determined.
This rearrangement is performed in the following order.
The position and order of the divided images to be selected for each frame are determined in descending order of importance within the group.
The first divided image is an I picture, the second is a P picture, and the third and subsequent pictures are B pictures.

  By doing this, it is possible to efficiently compress moving images while maintaining the reference relationship even when encoding with MPEG2.

  In the present embodiment, an apparatus for realizing the first and second embodiments will be described. FIG. 5 shows a configuration example.

  An image input unit 501 that can capture a high-definition moving image, an image analysis unit 502 that performs image analysis of the high-definition moving image captured by the image input unit 501, and an encoded moving image that temporarily buffers the high-definition moving image A buffer unit 503, a moving image encoding unit 504 that performs moving image encoding capable of dividing and encoding a moving image, a recording medium unit 505 that records a moving image stream, a recording medium control 506 that controls the recording medium unit, and image analysis The host unit 507 performs control of the encoder unit and control of the recording medium control unit based on the analysis result from the unit. In addition, a moving image decoder unit 508 that decodes the moving image stream encoded by the moving image encoding unit 504, a decoded moving image buffer unit 509 that temporarily buffers the decoded high-definition moving image, and a decoded moving image buffer unit 509 The image output unit 510 outputs a high-definition moving image to a monitor or the like. The host unit 507 controls reading of the recording medium control unit 506, the moving image decoder unit 508, and the decoded moving image buffer unit 509.

  At the time of image recording, the high-definition moving image captured by the image input unit 501 is analyzed for importance and correlation by the image analysis unit 502. The host unit 507 reads the analysis result from the image analysis unit 502, and determines the position and order of the divided images to be selected for each frame, the picture type, and the rate control bit assignment. Then, the host unit 507 sets the information in the moving image encoding unit 504, performs encoding, and stores the recording stream in the recording medium 506.

  At the time of image reproduction, the host unit 507 reads the recording stream from the recording medium 506, decodes it by the moving image decoder unit 508, and stores the decoded moving image in the decoding moving image buffer unit 509 based on the obtained image position information. The divided images are read out as one image from the decoded moving image buffer unit 509 and output to the image output unit 510.

Conceptual diagram of the present invention Moving image encoding method of embodiment 1 Moving image encoding method of embodiment 2 Quantization value control method when encoding divided images Example of moving image recording / playback device

Claims (5)

Dividing the screen into moving images, selecting one of the divided screens in order for each frame, compressing them by inter-image coding, and recording them together with the divided screen position information,
The recorded image data is expanded, and only the expanded screen of the divided screens for each frame is updated using the recorded divided screen position information, and the other divided screens are images that have been expanded in the past. A high-definition moving image recording / reproducing method characterized by holding and reproducing. In claim 1,
A high-definition moving image recording / reproducing method, wherein the number of images encoded per second of a divided image differs depending on the position of the divided screen. In claim 2,
A high-definition moving image recording / reproducing method, wherein the number of images encoded per second of a divided image is changed with respect to each divided screen position in accordance with image information of the moving image. In claim 3,
The image information of the moving image is a high-definition moving image recording characterized by a human region in the image, a motion vector, a dispersion value for each dispersed image, a difference value for each dispersed image, and a frequency conversion coefficient of the dispersed image. Playback method. An image input unit that captures high-definition moving images, an image analysis unit that performs image analysis of the high-definition moving images captured by the image input unit, an encoded moving image buffer unit that temporarily buffers high-definition moving images, and a moving image Based on the analysis result from the image analysis unit based on the analysis result from the image encoding unit, the moving image encoding unit that performs division encoding, the recording medium unit that records the moving image stream, the recording medium control that controls the recording medium unit A host unit that controls the control and the recording medium control unit, a moving image decoder unit that decodes the moving image stream encoded by the moving image encoding unit, and a decoded moving image buffer that temporarily buffers the decoded high-definition moving image And an image output unit that outputs the high-definition moving image of the decoded moving image buffer unit to a monitor or the like,
Dividing the screen into moving images, selecting one of the screens divided in each frame in order, compressing by inter-image coding, and recording with the divided screen position information,
The host unit controls the reading of the recording medium control unit, the moving image decoder unit, and the decoded moving image buffer unit, and uses the divided screen position information in which only the decoded screen is recorded out of the divided screens for each frame. The high-definition moving image recording / reproducing apparatus is characterized in that the other divided screens hold and reproduce the previously decoded images.

Images