JP2011109350A - Stereoscopic video encoder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To alleviate concentration of accesses on a shared memory for encoding. <P>SOLUTION: A left eye image encoding unit (101) encodes a left eye video signal by any one of in-picture encoding, inter-picture forward direction prediction encoding, and inter-picture bidirectional prediction encoding. A right eye image encoding unit (102) encodes a right eye video signal by any one of in-picture encoding, inter-picture forward direction prediction encoding, and inter-picture bidirectional prediction encoding. The left eye image encoding unit (101) and the right eye image encoding unit (102) share a memory 104 via a memory I/F 103. A picture type control unit (105) controls picture types of encoding in the left eye image encoding unit (101) and the right eye image encoding unit (102) so that timing of inter-picture bidirectional prediction encoding does not become simultaneous. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a stereoscopic video encoding apparatus.

JP 2002-252859 A

  2. Description of the Related Art Conventionally, in a twin-lens stereoscopic video TV, a left-eye image and a right-eye image captured from two different directions by two cameras are generated, and these are combined and displayed on the same screen to display a stereoscopic video. Display as. When such a stereoscopic video is recorded, if the left-eye image and the right-eye image are recorded respectively, the data amount is twice as much as that of the conventional art.

  As a method of reducing the entire data amount, compression coding using parallax compensation prediction using redundancy between viewpoints between a sub-image (for example, a right-eye image) and a main image (for example, a left-eye image) is performed. It has been proposed to do this (Patent Document 1). The main image is encoded by motion compensation prediction. Both the parallax compensation prediction and the motion compensation prediction can be applied to the sub-image, and an evaluation value of the prediction accuracy of the parallax compensation prediction and the motion compensation prediction is calculated, and the parallax compensation prediction or the motion compensation is calculated according to the evaluation value. Decide whether to predict.

  When encoding a left-eye image and a right-eye image of a binocular stereoscopic video, a process corresponding to twice the conventional method is required. When a memory for storing a reference image accompanying the encoding of each image is shared, the transmission bandwidth required for the memory also increases. When the inter-screen bi-directional predictive encoding is simultaneously processed for the left-eye image and the right-eye image, the memory access is most concentrated, and it is necessary to perform the memory access with a clock having a high frequency corresponding to the memory access. This is not preferable from the viewpoint of power consumption.

  It is an object of the present invention to provide a stereoscopic video encoding device that alleviates such inconvenience.

  A stereoscopic video encoding apparatus according to the present invention is a stereoscopic video encoding apparatus that encodes a left-eye video signal and a right-eye video signal, and that encodes a left-eye video signal. Means, a right-eye image encoding unit that encodes a right-eye video signal, a left-eye image encoding unit, and a right-eye image encoding unit, and stores data associated with the encoding. The picture type of the encoding in the memory means and the left-eye image encoding means and the right-eye image encoding means is set to intra-screen encoding, inter-screen forward prediction encoding, and inter-screen bidirectional prediction encoding. Picture type control means for controlling the picture type control means, wherein the picture type control means simultaneously performs inter-screen bi-directional predictive encoding timing in the left-eye image encoding means and the right-eye image encoding means. So that it does not become And controlling the Yataipu.

  According to the present invention, access to the memory means can be eased by controlling the picture type so that inter-screen bi-directional predictive encoding is not performed simultaneously in left-eye image encoding and right-eye image encoding. For example, it becomes possible to drive the memory with a clock having a lower frequency, and a reduction in power consumption in encoding can be expected.

1 is a block diagram of a schematic configuration of a stereoscopic video encoding apparatus according to a first embodiment of the present invention. It is a schematic block diagram of a left-eye image encoding unit (and a right-eye image encoding unit). It is a figure which shows the GOP structure of the image for left eyes, and the image for right eyes. It is a schematic block diagram of the second embodiment of the present invention. It is a figure which shows the example of an encoding timing of the image for left eyes and the image for right eyes in 2nd Example.

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

  FIG. 1 is a block diagram showing a schematic configuration of a stereoscopic video encoding apparatus according to an embodiment of the present invention. The left-eye image encoding unit 101 that encodes the left-eye video signal and the right-eye image encoding unit 102 that encodes the right-eye video signal share the memory 104 via the memory I / F 103. ing. Data accompanying encoding is temporarily stored in the memory 104. The memory I / F 103 mediates memory access from the left-eye image encoding unit 101 and the right-eye image encoding unit 102 and performs reading and writing to the memory 104. The left-eye image encoding unit 101 and the right-eye image encoding unit 102 are picture types in accordance with instructions from the picture type control unit 105 that controls the picture type in encoding, and the left-eye image and the right-eye image, respectively. Is encoded. In this embodiment, there are three types of picture types: an intra-picture coded picture (I picture), an inter-picture forward prediction coding picture (P picture), and an inter-picture bi-directional prediction coding (B picture).

  The left-eye image encoding unit 101 and the right-eye image encoding unit 102 correspond to the MPEG-4 AVC (ISO / IEC 14496-10) system and have the same configuration. FIG. 2 shows a schematic block diagram of the left-eye image encoding unit 101 and the right-eye image encoding unit 102. The image encoding unit illustrated in FIG. 2 performs processing in units of macroblocks obtained by dividing the encoding target screen into 16 × 16 pixel blocks.

  The prediction method determination unit 201 determines a prediction method for each macroblock in the encoding target screen according to the picture type instructed from the picture type control unit 105. Specifically, the prediction method determination unit 201 performs inter-screen prediction processing including simple intra-screen prediction or motion detection from the input video signal and the encoded pixel value read from the memory 104, and the encoding efficiency is improved. Determine the optimal prediction method. When the encoding target macroblock is an I slice, the intra prediction pixel block size and the prediction mode are determined. In the case of P slice or B slice, the one with higher coding efficiency is selected from among intra prediction or inter prediction. In the case of inter-screen prediction, intra-screen prediction encoding parameters such as an intra-screen prediction pixel block size and an intra-screen prediction mode are determined. In the case of inter-screen prediction, parameters for inter-screen prediction encoding such as a reference image frame, a macroblock division pattern, and a motion vector are determined.

  The prediction processing unit 202 generates a prediction image from the encoded image from the memory 104 according to the prediction encoding parameter specified by the prediction method determination unit 201, and outputs the prediction image to the local decoding unit 204. The prediction processing unit 202 also generates a prediction residual signal that is a difference between the encoding target image (pixel block) and the prediction image, and outputs the prediction residual signal to the orthogonal transform quantization unit 203.

  The orthogonal transform quantization unit 203 performs orthogonal transform processing by integer precision discrete cosine transform and discrete Hadamard transform in designated pixel block units (8 × 8 pixels or 4 × 4 pixel block units). The discrete Hadamard transform is applied to a direct current (DC) component obtained by a specific integer precision discrete cosine transform. The luminance component is applied to a direct current (DC) component when the in-screen prediction process is performed in units of 16 × 16 pixel blocks. The color difference signal is applied to a DC (direct current) component resulting from integer precision discrete cosine transform of each pixel block.

  The orthogonal transform quantization unit 203 quantizes the transform coefficient obtained by the orthogonal transform in a quantization step according to a designated quantization parameter, and the quantized transform coefficient data is entropy coding unit 205 and a local decoding unit. 204.

  The local decoding unit 204 performs inverse quantization and inverse orthogonal transform on the quantized transform coefficient data from the orthogonal transform quantization unit 203, adds the predicted image data from the prediction processing unit 202, and performs local decoding. The decoded image data is stored in the memory 104. The image data stored in the memory 104 is used for subsequent intra-screen prediction processing. Further, the image data that has been decoded and subjected to the deblocking filter processing is also stored in the memory 104 and used for the subsequent inter-screen prediction processing.

  The entropy encoding unit 205 performs entropy encoding on the quantized transform coefficient data from the orthogonal transform quantizing unit 203. Entropy coding includes context adaptive variable length coding (CAVLC). In addition, Context-based Adaptive Binary Arithmetic Coding (CABAC) may be employed. The encoded data by the entropy encoding unit 205 is supplied to the multiplexing processing unit 206.

  The multiplexing processing unit 206 multiplexes system data (such as a slice header) not shown in the encoded video data from the entropy encoding unit 205 and outputs the multiplexed data as encoded data.

  Among the I picture, P picture, and B picture, the B 104 picture that can refer to two pictures has the most memory access to the memory 104. For example, as shown in FIG. 3A, consider a case where the left-eye image and the right-eye image are encoded with the same GOP (Group Of Pictures) structure. Then, it is assumed that the data transmission amount accompanying the encoding of the P picture is twice the data transmission amount accompanying the encoding of the I picture and the data transmission amount accompanying the encoding of the B picture is four times the data transmission amount to the memory 104. To do. As shown in the lower graph of FIG. 3A, memory accesses at the B picture encoding processing timing are extremely concentrated.

  Therefore, the picture type control unit 105 controls the picture type with a GOP structure such that the encoding process using the B picture does not occur simultaneously in the left-eye image and the right-eye image. FIG. 3B shows an example of the GOP structure of the left eye image and the right eye image controlled by the picture type control unit 105. In this example, the P-picture is arranged in the third frame of the right-eye image, so that the B-picture encoding timing is not reached at the same time for the left-eye image and the right-eye image.

  As described above, by encoding with the GOP structure different between the left-eye image and the right-eye image, memory access in the encoding can be leveled. That is, as shown in the lower graph of FIG. 3B, it is possible to suppress the maximum transmission band required for the memory 104 with respect to FIG.

  FIG. 4 shows a schematic block diagram of a second embodiment of the present invention. The left-eye image encoding unit 401 that encodes the left-eye image and the right-eye image encoding unit 402 that encodes the right-eye image share the memory 404 via the memory I / F 403. . The memory I / F 403 mediates memory access from the left-eye image encoding unit 401 and the right-eye image encoding unit 402 and performs reading and writing to the memory 404. The left-eye image encoding unit 401 and the right-eye image encoding unit 402 are picture types in accordance with instructions from the picture type control unit 405 that controls the picture type in encoding, and the left-eye image and the right-eye image, respectively. Is encoded. In this embodiment, there are three types of picture types: an intra-picture coded picture (I picture), an inter-picture forward prediction coding picture (P picture), and an inter-picture bi-directional prediction coding (B picture).

  Details of the left-eye image encoding unit 401 and the right-eye image encoding unit 402 are the same as those of the left-eye image encoding unit 101 and the right-eye image encoding unit 102, and thus the description thereof is omitted.

  As will be described below, the picture type control unit 405 performs the left-eye image encoding unit 401 and the right-eye image so that the encoding process using the B picture does not occur simultaneously in the left-eye image and the right-eye image. The encoding unit 402 is controlled.

  FIG. 5 shows an example of timing for encoding the left-eye image and the right-eye image by the picture type control unit 405 when the left-eye image encoding and the right-eye image encoding have the same GOP structure. . In this example, the encoding timing of the right-eye image is delayed by one screen. As a result, the encoding timing of the B picture does not coincide with the left-eye image and the right-eye image. In addition to the delay for one screen, the encoding timing of one of the left-eye image coding and the right-eye image coding may be controlled to be delayed by a predetermined screen such as three screens or five screens. .

  In this way, the memory access in the encoding can be leveled by encoding the left-eye image and the right-eye image at different timings. As shown in the lower graph of FIG. 5, it is possible to suppress the maximum transmission bandwidth required for the memory 104 with respect to FIG.

101: Left-eye image encoding unit 102: Right-eye image encoding unit 104: Memory 105: Picture type control unit

Claims (3)

A stereoscopic video encoding device that encodes a left-eye video signal and a right-eye video signal,
Left-eye image encoding means for encoding the left-eye video signal;
Right-eye image encoding means for encoding a right-eye video signal;
Memory means for storing data associated with encoding, shared by the left-eye image encoding means and the right-eye image encoding means;
The picture type of encoding in the left-eye image encoding means and the right-eye image encoding means is controlled to any one of intra-frame encoding, inter-screen forward prediction encoding, and inter-screen bidirectional prediction encoding. And a picture type control means for
The picture type control means controls the picture type so that the timing of inter-picture bi-directional predictive encoding does not coincide in the left-eye image encoding means and the right-eye image encoding means. Stereoscopic video encoding device.   2. The picture type control means controls to make different GOP structures of encoded video data generated by the left-eye image encoding means and the right-eye image encoding means, respectively. The stereoscopic video encoding device described in 1. When the GOP structure of the encoded video data generated by the left-eye image encoding unit and the right-eye image encoding unit is the same, the picture type control unit is configured to perform the left-eye image encoding unit. 2. The stereoscopic video encoding apparatus according to claim 1, wherein one of the right-eye image encoding means is controlled to be delayed by a predetermined screen.

Images