JP6268715B2 - Moving picture encoding apparatus and program, moving picture decoding apparatus and program, and moving picture distribution system - Google Patents

Description

  The present invention relates to a moving image encoding device and program, a moving image decoding device and program, and a moving image distribution system. For example, the present invention relates to a moving image distribution system that encodes and distributes a moving image by a DVC (Distributed Video Coding) method. Applicable.

  A system that adopts a conventional DVC (Distributed Video Coding) system is a system that encodes and decodes a moving picture based on the Slepian-Wolf theory and the Wyner-Ziv theory.

  In the DVC method, a code (hereinafter referred to as “Wyner-Ziv code”) that reconstructs an encoding target image from a prediction image (hereinafter referred to as “decoder predicted image”) of the encoding target image generated by the decoder. The decoder prediction image is generated without referring directly. Due to this feature, a moving picture encoding apparatus adopting the DVC method does not need to include a complicated predicted image generation unit, and the amount of calculation required for encoding can be reduced. The problem with the DVC scheme is that it is difficult to obtain a Wyner-Ziv code rate that is necessary and sufficient to generate a decoded image while retaining the decoder predicted image itself. For example, in a video encoding apparatus adopting the DVC method, if there are too many Wyner-Ziv codes, a compression effect cannot be obtained, and if there are too few Wyner-Ziv codes, a decoded picture cannot be generated from the predicted decoder picture.

  As a conventional technique for solving such a problem of the DVC method, there is a technique described in Non-Patent Document 1. In Non-Patent Document 1, a method for searching for a necessary and sufficient rate by continuously requesting (feeding back) an additional Wyner-Ziv code until a decoded image is generated from a decoder predicted image (hereinafter referred to as “decoder rate control method”). Called).

  However, the conventional decoder rate control method has a disadvantage in that the application is limited because it cannot be encoded without the presence of a decoding process via a network. And as a prior art for improving this shortcoming, there are technologies described in Non-Patent Documents 2 and 3.

  In order to eliminate this shortcoming, in Non-Patent Document 2 and Non-Patent Document 3, an encoder predicted image (a predicted image that can be generated without requiring a calculation as complicated as a decoder predicted image) is generated on the moving image encoding device side. A method of controlling the rate by comparing the encoding target image and the predicted encoder image (hereinafter referred to as “encoder rate control method”) has been proposed.

  In the moving picture coding apparatus adopting the encoder rate control method of Non-Patent Documents 2 and 3, by performing rate control, the rate of the Wyer-Ziv code sent for the first time is brought close to a necessary and sufficient rate, and the number of feedbacks is set. Less.

B. Girod, a M.M. Aaron, S.A. Rane, andD. Rebolo-Monedero, “Distributed Video Coding,” Proceedings of the IEEE, vol. 93, Jan. 2005, pp. 71-83. C. Brites and F.M. Pereira, “Encoder rate control for domain domain Wyner-Ziv video coding,” 1 image Processing 2007. ICIP 2007. IEEE International Conference on, IEEE, 2007, pp. 4-7. M.M. Tagliasacchi, A .; Majmardar, and R.M. Ramchandran, “A distributed-source-coding based robust-spatial-temporal scalable video codec,” Proc. Picture Coding Symposium, Citeseeer, 2004. J. et al. Ascenso, C.I. Brites, and F.B. Pereira, 'Motion compensated refining for low complexity pixel based distributed video coding, "Proceedings. IEEE Conference on Advanced.

  When an encoder predicted image (side information) is generated by a conventional moving image encoding device, according to the amount of error between the quantized value of the generated encoder predicted image and the quantized value of the Wyner-Ziv image, Increase or decrease the amount of code supplied to the video decoding device side. That is, in the conventional video encoding device, the larger the amount of quantization value errors described above, the larger the amount of code bits supplied to the video decoding device side.

  On the other hand, in the video encoding device, it is preferable from the viewpoint of transmission efficiency and the like that the code amount is smaller, but if the code amount is reduced, the quality of the decoded image generated on the video decoding device side is lowered.

  Therefore, a moving image encoding device and program, a moving image decoding device and program, and a moving image distribution system that can reduce the amount of code without degrading the quality of the decoded image are desired.

According to a first aspect of the present invention, there is provided a moving image encoding apparatus for encoding a moving image having a frame sequence. (1) A predicted image that generates a predicted image of a non-key frame using a key frame in the frame sequence. A generating unit; (2) a complexity calculating unit that calculates the degree of complexity of a non-key frame; and (3) a bit depth related to quantization of each parameter constituting an image based on a calculation result of the complexity calculating unit. And (4) quantizing each parameter of the non-key frame and each parameter of the predicted image with the bit depth determined by the bit depth determining means, For generating a frame and a predicted image after quantization, for a parameter for which a bit depth smaller than the maximum bit depth is determined, Data obtained by adding padding bits of a predetermined pattern corresponding to the difference from the most bit depth to either the most significant bit side or the least significant bit side of the quantized data corresponding to the specified bit depth And (5) error correction code generation means for generating an error correction code for correcting an error in the predicted image after quantization with respect to the non-key frame after quantization, (6) ) parameters of each image processed by the bit depth determination means and the quantizing means, the pixel values of the pixels constituting the image, or transform coefficient obtained by converting the image by a predetermined conversion method It is characterized by a quantized value for each of several regions.

According to a second aspect of the present invention, in a moving image decoding apparatus for decoding moving image data in which a moving image having a frame sequence is encoded in units of frames, (1) key frame holding means for holding key frame data; 2) predicted image generation means for generating a predicted image of a non-key frame in the frame sequence using the key frame held by the key frame holding means; and (3) a non-key frame constituting the moving image data. Bit depth holding means for holding bit depth information related to the bit depth applied to quantization of each parameter constituting the image in the process of generating the encoded data of (4), and (4) held by the bit depth holding means The bit depth is used to quantize each parameter of the predicted image to generate a quantized predicted image, which is smaller than the maximum bit depth. The parameter for which the bit depth is determined corresponds to the difference from the most bit depth on either the most significant bit side or the least significant bit side of the quantized data for the bit depth set for the parameter. Quantization means for generating data to which padding bits of a predetermined pattern are added as quantized data of the parameter; and (5) encoded data of non-keyframes constituting the moving image data includes a predicted image after quantization. In the case of an error correction code for correction, error correction means for generating a corrected image obtained by correcting the quantized predicted image using the error correction code, (6) the corrected image, and the predicted image And (7) the bit depth determination unit and the quantization unit, the reconstruction unit configured to generate a reconstructed image based on the reconstruction of the non-keyframe image. Parameters of each image processed by the step, the pixel values of the pixels constituting the image, or it is quantized value for each conversion coefficient region obtained by converting the image by a predetermined conversion method It is characterized by.

A moving image encoding program according to a third aspect of the present invention provides a computer mounted on a moving image encoding apparatus that encodes a moving image having a frame sequence, using (1) a key frame in the frame sequence, A predicted image generating means for generating a predicted image of a non-key frame; (2) a complexity calculating unit for calculating a complexity level of the non-key frame; and (3) an image based on a calculation result of the complexity calculating unit. Bit depth determining means for determining a bit depth relating to quantization of each parameter constituting; (4) each bit parameter determined by the bit depth determining means, each parameter of the non-key frame, and each prediction image The parameter is quantized to generate a non-key frame after quantization and a predicted image after quantization, and a bit depth that is less than the maximum bit depth is determined. For the meter, padding bits of a predetermined pattern corresponding to the difference from the most bit depth are placed on either the most significant bit side or the least significant bit side of the quantized data for the bit depth determined for the parameter. Quantization means for generating the added data as quantized data of the parameter; and (5) an error correction code for generating an error correction code for correcting an error in the predicted image after quantization for the non-key frame after quantization. (6) The parameter of each image processed by the bit depth determination unit and the quantization unit is the pixel value of each pixel constituting the image, or the image is converted by a predetermined conversion method. characterized in that it is a quantized value for each conversion coefficient region obtained by converting.

A moving image decoding program according to a fourth aspect of the present invention provides a computer mounted on a moving image decoding apparatus that decodes moving image data in which a moving image having a frame sequence is encoded in units of frames. (2) a predicted image generating unit that generates a predicted image of a non-key frame in the frame sequence using the key frame held by the key frame holding unit; and (3) Bit depth holding means for holding bit depth information related to bit depth applied to quantization of each parameter constituting an image in the process of generating encoded data of non-key frames constituting the moving image data; (4 ) The quantized prediction image is generated by quantizing each parameter of the prediction image with the bit depth held by the bit depth holding means. For a parameter for which a bit depth less than the maximum bit depth is determined, either the most significant bit side or the least significant bit side of the quantized data corresponding to the bit depth set for the parameter is set. , A quantization means for generating, as quantized data of the parameter, data added with a predetermined pattern of padding bits corresponding to the difference from the most bit depth, and (5) a non-key frame of the moving image data If the encoded data is an error correcting code for correcting the quantized predicted image, an error correcting means for generating a corrected image obtained by correcting the quantized predicted image using the error correcting code; (6 ) Based on the corrected image and the predicted image, it functions as a reconstruction unit that generates a reconstructed image obtained by reconstructing a non-keyframe image. (7) The parameters of each image processed by the bit depth determining means and the quantizing means are obtained by converting the pixel value of each pixel constituting the image or converting the image by a predetermined conversion method. characterized in that it is a quantized value for each conversion coefficient region.

  According to a fifth aspect of the present invention, there is provided a moving image encoding device for generating moving image data obtained by encoding a moving image having a frame sequence in units of frames, and a moving image for decoding the moving image data generated by the moving image encoding device. In a moving image distribution system including an image decoding device, (1) the moving image encoding device according to the first aspect of the present invention is applied as the moving image encoding device, and (2) the second image as the moving image decoding device. The moving picture decoding apparatus of the invention is applied.

  ADVANTAGE OF THE INVENTION According to this invention, the moving image delivery system which can reduce code amount, without reducing the quality of a decoded image can be provided.

It is the block diagram shown about the whole structure of the moving image delivery system which concerns on 1st Embodiment. It is the block diagram shown about the functional structure of the moving image encoder which concerns on 1st Embodiment. It is the block diagram shown about the functional structure of the moving image decoding apparatus which concerns on 1st Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image encoder side) of the moving image delivery system which concerns on 1st Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image decoding apparatus side) of the moving image delivery system which concerns on 1st Embodiment. It is explanatory drawing shown about the complexity produced | generated with the moving image encoder which concerns on 1st Embodiment, and the content of the bit depth control signal. It is explanatory drawing shown about the example of the quantization process of the image performed with the moving image encoder which concerns on 1st Embodiment. It is the block diagram shown about the functional structure of the moving image encoder which concerns on 2nd Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image encoder side) of the moving image delivery system which concerns on 2nd Embodiment. It is the block diagram shown about the functional structure of the moving image encoder which concerns on 3rd Embodiment. It is the block diagram shown about the functional structure of the moving image decoding apparatus which concerns on 3rd Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image encoder side) of the moving image delivery system which concerns on 3rd Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image decoding apparatus side) of the moving image delivery system which concerns on 3rd Embodiment. It is the block diagram shown about the functional structure of the moving image encoder which concerns on 4th Embodiment. It is the block diagram shown about the functional structure of the moving image decoding apparatus which concerns on 4th Embodiment. It is the flowchart shown about operation | movement (operation | movement by the moving image encoder side) of the moving image delivery system which concerns on 4th Embodiment. It is the flowchart shown about the operation | movement (operation | movement by the moving image decoding apparatus side) of the moving image delivery system which concerns on 4th Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a moving image encoding apparatus and program, a moving image decoding apparatus and program, and a moving image distribution system according to the present invention will be described in detail with reference to the drawings. .

(A-1) Configuration of First Embodiment The moving image distribution system 1 includes a moving image encoding device 10 and a moving image decoding device 20.

  FIG. 1 is a block diagram showing the overall configuration of a moving image distribution system 1 of this embodiment.

  The moving image distribution system 1 includes a moving image encoding device 10 and a moving image decoding device 20.

  The moving image encoding device 10 encodes an input moving image signal (input video signal) input in frame units (image units), and streams the encoded data (bit stream) to output. As shown in FIG. 1, in the moving image encoding device 10, each image constituting the input moving image signal is encoded by being divided into a Wyner-Ziv image F11 and a Key image F12. Then, the moving image encoding apparatus 10 encodes a stream of encoded data obtained by encoding the Wyner-Ziv image F11 (hereinafter referred to as “Slepian-Wolf stream ST1”), and a stream of data obtained by encoding the Key image F12 ( Hereinafter, it is referred to as “Key stream ST2”) and a bit depth transfer code C1 (described later in detail) is output.

  The video decoding device 20 decodes the encoded data (Slepian-Wolf stream and Key stream) output from the video encoding device 10, generates a decoded image (decoded frame), and generates a frame unit (image unit). ) Is output as a decoded moving image signal (decoded moving image data). The moving picture decoding apparatus 20 combines a decoded moving picture signal obtained by decoding a Wyner-Ziv decoded picture F41 obtained by decoding the Slepian-Wolf stream ST1 and a decoded picture (Key picture F12) obtained by decoding the Key stream ST2. Is output.

  Note that the data format output by the video encoding device 10 and the data format input by the video decoding device 20 are not limited. For example, instead of the stream format suitable for real-time data transmission from the moving image encoding device 10, similar data is output as one data file, and the moving image decoding device is offline (for example, a data recording medium such as a hard disk). 20 may be supplied.

  Next, the internal configuration of the moving picture coding apparatus 10 will be described with reference to FIG.

  The moving image encoding apparatus 10 includes a bit depth control unit 101, an encoding target image quantization unit 104, an encoder predicted image generation unit 105, an encoder predicted image quantization unit 106, a rate control unit 107, and a Slepian-Wolf encoding unit 108. , A bit depth notification unit 109 and a key image encoding unit 120. The bit depth control unit 101 includes a complexity calculation unit 102 and a control unit 103.

  The Key image encoding unit 120 performs an intra-screen encoding process (for example, an encoding process using an encoding technique such as H.264 / AVC or JPEG), encodes the Key image F12, and streams the encoded data. And output as a Key stream ST2.

  The encoder predicted image generation unit 105 generates an encoder predicted image F21 from the Key image F12. The method by which the encoder predicted image generation unit 105 generates the encoder predicted image F21 is not limited, and various generation methods in a system employing the DVC system can be applied.

  The encoder predicted image quantization unit 106 quantizes the encoder predicted image F21 (for example, converts it into bit string data for each pixel), and generates a quantized encoder predicted image F22.

  The rate control unit 107 estimates a code amount necessary for the Slepian-Wolf code from the quantized encoder predicted image F22 and the quantized Wyner-Ziv image F23, and outputs the estimation result as a rate R. The specific process in which the rate control unit 107 estimates the code amount is not limited, and various estimation processes in a system employing DVC can be applied.

  The Slepian-Wolf encoding unit 108 performs Slepian-Wolf encoding on the quantized Wyner-Ziv image F23, and generates a Slepian-Wolf stream ST1 with a code amount based on the rate R. The specific method in which the Slepian-Wolf encoding unit 108 performs the Slepian-Wolf encoding process is not limited, and various encoding processes in a system employing DVC can be applied.

  The encoding target image quantizing unit 104 is a quantized Wyner-Ziv image F23 based on the Wyner-Ziv image F11, the encoder predicted image F21, and the control of the bit depth notifying unit 109 (a bit depth control signal S12 described later). Is generated.

  The bit depth control unit 101 divides a frame into blocks of an arbitrary size, and calculates a parameter (bit depth described later) used for image quantization for each block constituting the Wyner-Ziv image F11. And a complexity calculation unit 102 and a control unit 103.

  The complexity calculation unit 102 divides the frame (Wyner-Ziv image F11) into blocks of a predetermined size (in this embodiment, it is assumed to be divided into 4 × 4 blocks as an example), and Wyner-Ziv. The complexity is calculated for each block constituting the image F11. The complexity calculator 102 outputs the complexity obtained for each block as the complexity S11.

  The method of calculating the complexity by the complexity S11 is not limited. For example, the complexity may be calculated based on the size of the AC component for each block, the strength of the edge, or the like.

  First, an example in which the complexity S11 calculates the complexity of each block based on the AC component for each block will be described. For example, the complexity calculation unit 102 compares the average value of the pixel values in each block of the Wyner-Ziv image F11 with the average value of the pixel values of all the blocks (the pixel values of the entire image), and calculates the difference for each block. It may be determined that the AC component of the block is larger (that is, the complexity is higher) as is larger. Further, for example, when the moving image encoding apparatus 10 includes a scalable encoding mechanism as in the technique described in Non-Patent Document 4, the complexity calculation unit 102 generates the Wyner-Ziv image F11 generated by the scalable encoding mechanism. An enlarged image obtained by enlarging the reduced image may be generated, and the complexity S11 may be calculated by evaluating the magnitude of the signal of the difference image between the Wyner-Ziv image F11 and the enlarged image. Further, the complexity calculation unit 102 regards the reduced image as DC component information, calculates the AC component by subtracting it from the Wyner-Ziv image F11, and calculates the complexity S11 by evaluating the size of the AC component. You may make it do.

  Next, an example in which the complexity S11 calculates the complexity of each block based on the edge strength for each block will be described. For example, the complexity calculation unit 102 first uses a filter (for example, a filter such as a Sobel filter) for detecting an edge in the entire Wyner-Ziv image F11 to determine whether or not the pixel belongs to the edge portion. To detect. The complexity calculation unit 102 counts the number of pixels belonging to the edge portion (hereinafter referred to as “edge pixels”) for each block, and the block having a larger number of edge pixels has a stronger edge (that is, the complexity is higher). It may be determined that the block is high.

  In this embodiment, in order to simplify the description, the complexity calculated by the complexity S11 for each block is indicated by a numerical value of 0 or 1. In other words, in this embodiment, the complexity calculated by the complexity S11 for each block is assumed to be expressed in two steps.

  For example, the complexity calculation unit 102 calculates one or a plurality of parameters (for example, an AC component and / or edge strength) as a basis of complexity for each block, and the complexity of the block is calculated based on the calculated parameters. The degree (0 or 1) may be determined. For example, the complexity calculation unit 102 compares a predetermined threshold value k with the AC component AC of each block, sets the complexity of the block to 1 when AC ≧ k, and sets the complexity of the block to 0 otherwise. You may make it.

  Based on the complexity S11 supplied from the complexity calculation unit 102, the control unit 103 uses, for each block, the number of bits for each pixel used for quantization of the Wyner-Ziv image F11 (hereinafter referred to as “bit depth”). )). And the control part 103 outputs the bit depth calculated | required for every block as the bit depth control signal S11.

  For example, when the complexity S11 is high, the bit depth control signal S12 is considered to have high prediction difficulty in the predicted image, and the bit depth control signal S12 is generated so as to be encoded with a high bit depth. Is low, the bit depth control signal S12 is generated so that the prediction difficulty of the predicted image is low and the coding is performed with a low bit depth.

  The complexity S11 of this embodiment is represented by 0 or 1 as described above. In this embodiment, each pixel value of the Wyner-Ziv image F11 is represented by an integer from 0 to 255 (decimal number). That is, each pixel value of the Wyner-Ziv image F11 is represented by an 8-digit binary number (8-bit data).

  Therefore, in this embodiment, the control unit 103 sets the bit depth of the block in which the complexity is set to 0 in the complexity S11 to 3, and the bit depth of the block in which the complexity is set to 1 in the complexity S11 to 5. It will be described as being determined.

  Next, the internal configuration of the video decoding device 20 will be described with reference to FIG.

  The video decoding device 20 includes a bit depth receiving unit 200, a decoder predicted image generation unit 201, a decoder predicted image quantization unit 202, an LLR estimation unit 203, a Lepian-Wolf decoding unit 204, a reconstruction unit 205, and a Key image decoding unit 220. have.

  The key image decoding unit 220 decodes the key stream ST2 (decoding process corresponding to the moving image encoding device 10 side) to obtain a key image F12. In this embodiment, in order to simplify the description, the key image decoding unit 220 restores the key stream ST2, and a restored image having the same content as the key image F12 input to the encoding side is obtained. To do. The decoded image obtained by the key image decoding unit 220 is preferably as close as possible to the key image input to the encoding side, but may not be completely the same.

  The bit depth receiving unit 200 restores and acquires the bit depth control signal S11 from the bit depth transmission code C1 supplied from the moving image encoding device 10.

  The predicted decoder image generation unit 201 generates a predicted decoder image F31 from the Key image F12 acquired by the Key image decoding unit 220. The method by which the decoder predicted image generation unit 201 generates the decoder predicted image F31 is not limited, and various generation methods in a system employing the DVC method can be applied.

  The decoder predicted image quantization unit 202 quantizes the decoder predicted image F31 (for example, converts it into bit string data for each pixel), and generates a quantized decoder predicted image F32.

  The LLR estimation unit 203 estimates an LLR (Log Likelihood Ratio) for each bit of the bit string quantized from the quantized decoder predicted image F32, and outputs the LLR notification signal S22.

  The Slepian-Wol decoding unit 204 generates a corrected Wyner-Ziv image F33 from the LLR notification signal S22 and the Slepian-Wolf stream ST1.

  In the moving image encoding device 10, the LLR estimation unit 203 and the Slepian-Wol decoding unit 204 form a decoding unit that performs Slepian-Wolf decoding. And as a concrete process which the LLR estimation part 203 and the Slepian-Wol decoding part 204 perform, the decoding means in various DVC systems is applicable.

  The reconstruction unit 205 uses the corrected Wyner-Ziv image F33 and the decoder predicted image F31 to perform image reconstruction processing (processing for estimating the Wyner-Ziv image F11 while performing inverse quantization), and Wyner-Ziv. The decoded image F41 is generated. Detailed processing of the reconstruction unit 205 will be described later.

(A-2) Operation of the First Embodiment Next, the operation of the moving image distribution system 1 of the first embodiment having the above configuration will be described.

(A-2-1) Operation of Moving Image Encoding Device 10 First, the operation on the moving image encoding device 10 side of the moving image distribution system 1 will be described with reference to the flowchart of FIG.

  First, it is assumed that the complexity calculation unit 102 calculates the complexity S11 for each block of the Wyner-Ziv image F11 and outputs the calculation result as the complexity S11 (S101).

  The control unit 103 calculates the bit depth of each block based on the complexity S11, and outputs the calculation result as the bit depth control signal S12 (S102).

  FIG. 6 is an example conceptually illustrating the contents of the complexity S11 and the bit depth control signal S12. In FIG. 6, the Wyner-Ziv image F11 is divided into 4 vertical blocks × 4 horizontal blocks (16 blocks). FIGS. 6A and 6B conceptually show the contents of the complexity S11 and the bit depth control signal S12, respectively. In FIG. 6, each square frame is a block, and the parameters (complexity or bit depth) corresponding to the block are included in the square.

  In FIG. 6, for the leftmost column, blocks B1, B2, B3, and B4 are assigned in order from the upper block. In FIG. 6A, the complexity of the blocks B1 to B4 is 1, 0, 0, and 1, respectively. Further, in FIG. 6B, the bit depths of the blocks B1 to B4 are bit depths corresponding to the complexity of FIG. 6A, respectively. Specifically, in FIG. 6B, the bit depths of the blocks B1 to B4 are 5, 3, 3, and 5, respectively.

  In the encoding target image quantization unit 104, the Wyner-Ziv image F11 is quantized using the bit depth of each block set in the bit depth control signal S12, and a post-quantization Wyner-Ziv image F23 is generated. (S103).

  FIG. 7 shows an example of quantization processing based on the bit depth control signal S12 by the encoding target image quantization unit 104.

  FIGS. 7A to 7D show examples of quantizing one pixel constituting each of the blocks B1 to B4 shown in FIG.

  The encoding target image quantization unit 104 first converts the pixel value (decimal number from 0 to 255) of each block constituting the Wyner-Ziv image F11 into an 8-bit binary number, and starts from the MSB (most significant bit). Bit data for bit depth is acquired as a quantized value. For example, in FIG. 7A, when the pixel value of an arbitrary pixel constituting the block B1 is converted to a binary number, it is “10111011”, and the bit depth of the block B1 is 5. Therefore, the encoding target image quantization unit 104 acquires “10111”, which is data of 5 bits, from the MSB as the quantization value of the pixel.

  However, when the bit depth is different in each block, the encoding target image quantization unit 104 of this embodiment outputs bit data having the number of digits corresponding to the longest bit depth for all blocks. . For example, in the example of FIG. 6, in the bit depth control signal S12, blocks having a bit depth of 3 and 5 are mixed. Therefore, the encoding target image quantization unit 104 outputs bit data normalized to 5 which is the longest bit depth as a quantized value. In this case, bit data that is insufficient for normalization is generated in a bit depth block that is shorter than the longest bit depth block. However, the encoding target image quantizing unit 104 arbitrarily selects bit data that is insufficient as bit data. The dummy data (hereinafter also referred to as “padding bits”) of the above pattern (hereinafter also referred to as “symbol”) is embedded. In this embodiment, the encoding target image quantization unit 104 will be described as applying a bit pattern that is all “1” as symbols used for padding bits.

  For example, in FIG. 7B, when the pixel value of an arbitrary pixel constituting the block B2 is converted into a binary number, it is “00101101”, and the bit depth of the block B2 is 3. However, since the maximum bit depth of the entire image is 5, the encoding target image quantization unit 104 acquires bit data corresponding to bit depth (data corresponding to 3 bits) in order from the MSB. Bit data to which padding bit “11” is added is acquired as a quantized value.

  Although the bit pattern used as a symbol is arbitrary, it is necessary to use a common symbol also in other quantization units in the moving image distribution system 1. In FIG. 7, padding bits are embedded on the LSB side, but may be embedded on the MSB side. However, the padding bit embedding location also needs to coincide with other quantization units and reconstruction units in the moving image distribution system 1.

  Then, the encoder predicted image generation unit 105 generates an encoder predicted image F21 from the key image F12 using motion estimation or the like (S104).

  In the encoder predicted image quantization unit 106, the encoder predicted image F21 is quantized using the bit depth control signal S12, and a quantized encoder predicted image F22 is generated (S105). Note that the method of quantization processing in the encoder predicted image F21 is the same as that of the encoding target image quantization unit 104, and thus detailed description thereof is omitted.

  In the rate control unit 107, the quantized Wyner-Ziv image F23 and the quantized encoder predicted image F22 are compared, and the rate R of the Slepian-Wolf encoding required for decoding by the video decoding device 20 (decoder). Is generated (S106).

  In the Slepian-Wolf encoding unit 108, the quantized Wyner-Ziv image F23 is subjected to Slepian-Wolf encoding based on the rate R, and a Slepian-Wolf stream ST1 is generated (S107).

  In the bit depth notification unit 109, the bit depth control signal S12 is converted into a bit depth transfer code C1 and output (S108). Since the bit depth control signal S12 is information in units of blocks, the bit depth transmission code C1 has a small amount of code compared to the image data.

(A-2-2) Operation of Video Decoding Device 20 Next, the operation of the decoding process of the video decoding device 20 will be described using the flowchart of FIG.

  First, the bit depth receiving unit 200 restores the bit depth control signal S11 based on the input bit depth transmission code C1 (S109).

  Then, the decoder predicted image generation unit 201 generates a decoder predicted image F31 from the Key image F12 using motion estimation or the like (S110).

  The decoder predicted image quantization unit 202 quantizes the decoder predicted image F31 using the bit depth control signal S12, and generates a quantized decoder predicted image F32 (S111). Note that the method of quantization processing in the decoder predicted image quantization unit 202 is the same as that of the encoding target image quantization unit 104, and thus detailed description thereof is omitted.

  The LLR estimation unit 203 estimates an LLR from the quantized decoder predicted image F32, and outputs an LLR notification signal S22 based on the estimated LLR (S112).

  In the Slepian-Wol decoding unit 204, Slepian-Wolf decoding is performed using the LLR notification signal S22 and the Slepian-Wolf stream ST1, and a corrected Wyner-Ziv image F33 is generated (S113).

  Finally, the reconstruction unit 205 generates a Wyner-Ziv decoded image F41 based on the bit depth control signal S12, the corrected Wyner-Ziv image F33, and the predicted decoder image F31 (S114).

  As described above, the padding bit embedding position based on the bit depth control signal S12 is common in each quantization unit constituting the moving image distribution system 1. Therefore, the reconstruction unit 205 determines the position of the padding bit included in the corrected Wyner-Ziv image F33 (that is, the quantized value of the pixel belonging to the block including the padding bit) based on the bit depth control signal S12. Can be recognized. Accordingly, the reconstruction unit 205 can perform reconstruction processing in which padding bits are ignored (discarded) in the corrected Wyner-Ziv image F33.

(A-3) Effects of First Embodiment According to the first embodiment, the following effects can be achieved.

  The moving image coding apparatus 10 regards a region having high complexity as low quality of the prediction image (encoder prediction image F21, decoder prediction image F31), and effectively allocates many error correction codes (a large bit depth is set). Can be assigned). Thereby, the encoding efficiency in the moving image encoder 10 and the decoding quality in the moving image decoder 20 can be improved.

  In addition, the moving image encoding device 10 and the moving image decoding device 20 use padding bits to fill in gaps between quantized values of different bit depths, so that the block size and the processing unit of the Slepian-Wolf code (how many bits There is no dependency between each of them (Slipian-Wolf encoding or decoding), and the block size can be easily changed according to the content. For example, in the case of content that changes at low pixel intervals with low complexity, even if the code amount of the bit depth transmission signal increases, it is more efficient to manually process by reducing the block size in the moving image distribution system 1. There is a case.

(B) Second Embodiment Hereinafter, a second embodiment of a moving image encoding apparatus and program, a moving image decoding apparatus and program, and a moving image distribution system according to the present invention will be described in detail with reference to the drawings. .

(B-1) Configuration of Second Embodiment The overall configuration of the moving image distribution system 1A of the second embodiment can also be shown using FIG. Hereinafter, the difference between the second embodiment and the first embodiment will be described.

  In the moving image distribution system 1A of the second embodiment, the moving image encoding device 10 is replaced with the moving image encoding device 10A.

  FIG. 8 is a block diagram illustrating a functional configuration of the moving image encoding device 10A according to the second embodiment.

  In the moving image encoding device 10A of the second embodiment, the bit depth control unit 101 is replaced with the bit depth control unit 101A. In the bit depth control unit 101A, the control unit 103 is replaced with the control unit 103A, and a motion detection unit 109 is further added.

  The motion detection unit 109 determines the presence or absence of motion between the Wyner-Ziv image F11 and the Key image F12 for each block (for each block divided in the same manner as the complexity calculation unit 102), and based on the determination result The motion information S13 is generated. The motion information S13 includes information on the presence / absence of motion for each block.

  The motion information S13 has a size of a cost function such as SAD (sum of absolute difference, sum of absolute difference) or MAD (mean absolute difference) when the motion vector is set to zero. It may be determined that there is a motion when the threshold is equal to or greater than a predetermined threshold, and it is determined that there is no motion in other cases. Further, when the moving image encoding apparatus 10 includes a scalable encoding mechanism as described in Non-Patent Document 4, it corresponds to the Wyner-Ziv image F11 and the Key image F12 generated by the scalable encoding mechanism, respectively. The presence / absence of motion for each block may be calculated by comparing the reduced images.

  The control unit 103A is different from the first embodiment in that the bit depth for each block is determined in consideration of the motion information S13 in addition to the complexity S11.

(B-2) Operation | movement of 2nd Embodiment Next, operation | movement of 1 A of moving image delivery systems of 2nd Embodiment which has the above structures is demonstrated.

  Hereinafter, differences from the first embodiment among the operations of the moving image distribution system 1A according to the second embodiment will be described with reference to FIG.

  The moving image distribution system 1A according to the second embodiment is different in that the above-described step S102 is replaced with S202.

  FIG. 9A is a flowchart showing the operation of the moving image encoding apparatus 10A in step S202 described above.

  In step S202-1, the motion detection unit 109 determines the presence / absence of motion for each block from the Wyner-Ziv image F11 and the Key image F12, and generates motion information S13 based on the determination result.

  In step S202-2, the control unit 203 determines a bit depth for each block based on the complexity S11 and the motion information S13, and generates a bit depth control signal S12 based on the determination.

  For example, when the motion information S13 suggests that there is little motion, it is considered that the prediction image is difficult to predict regardless of the complexity S11, and the bit is encoded so as to be encoded with a low bit depth. When the depth control signal 203 is generated and the motion information S13 indicates that the motion is large, the bit depth control signal 203 is generated according to the height of the complexity S11.

  FIG. 9B is an explanatory diagram showing rules for determining the bit depth of each block by the control unit 103A.

  In this embodiment, the control unit 103A has a high degree of complexity in the complexity S11, as shown in FIG. 9B, for a block in which no motion (less motion) is indicated in the motion information S13. Regardless of this, it is assumed that the prediction difficulty of the predicted image is low, and a small bit depth (3) is determined.

  Further, the control unit 103A applies a bit depth corresponding to the height of the complexity S11, as shown in FIG. 9B, for the block in which motion is present (the motion is large) in the motion information S13. Assume that the bit depth is determined (by the same method as in the first embodiment).

(B-3) Effects of Second Embodiment According to the second embodiment, the following effects can be achieved as compared with the first embodiment.

  In the moving image coding apparatus 10A, the quality of the predicted image (encoder predicted image F21, decoder predicted image F31) is low even if the region with little motion (the region without motion) is a highly complex region. Based on the assumption that it is high, an area with little motion is encoded with a small bit depth. Since the above hypothesis can be applied to many moving pictures, the control of the bit depth in the moving picture coding apparatus 10A becomes more appropriate (appropriate), and as a result, the coding efficiency and the decoding quality are further improved. Can do.

(C) Third Embodiment Hereinafter, a third embodiment of a moving image encoding device and program, a moving image decoding device and program, and a moving image distribution system according to the present invention will be described in detail with reference to the drawings. .

(C-1) Configuration of Third Embodiment The overall configuration of the moving image distribution system B of the third embodiment can also be shown using FIG. Hereinafter, the difference between the third embodiment and the second embodiment will be described.

  In the moving image distribution system B of the third embodiment, the second embodiment is that the moving image encoding device 10A and the moving image decoding device 20 are replaced with the moving image encoding device 10B and the moving image decoding device 20B. Is different.

  FIG. 10 is a block diagram illustrating a functional configuration of a moving image encoding device 10B according to the third embodiment.

  In the video encoding device 10B of the third embodiment, the rate control unit 107 is replaced with the rate control unit 107B.

  The rate control unit 107B specifies the position of padding bits based on the bit depth control signal S12, the quantized encoder predicted image F22, and the quantized Wyner-Ziv image F23, and targets only bits other than the padding bits. Then, a bit inversion probability is obtained, and a process for generating a rate R from the obtained bit inversion probability is performed.

  FIG. 11 is a block diagram illustrating a functional configuration of a video decoding device 20B according to the third embodiment.

  The moving picture decoding apparatus 20B according to the third embodiment is different from the first embodiment in that an LLR correction unit 209 is added.

  The LLR correction unit 209 clarifies the position of the padding bit from the bit depth control signal S11 and corrects the size of the LLR corresponding to the padding bit so that it is sufficiently larger than the LLRs of the other bits (for example, The corrected LLR is obtained by performing a process of correcting by multiplying one or more coefficients or a process of adding a constant. Then, the LLR correction unit 209 generates a post-correction LLR notification signal S23 representing the post-correction LLR of each bit, and supplies the generated LLR notification signal S23 to the Slepian-Wolf decoding unit 204.

  The Slepian-Wol decoding unit 204 performs Slepian-Wolf decoding using the corrected LLR notification signal S23 and the Slepian-Wolf stream ST1, and generates a corrected Wyner-Ziv image F33.

(C-2) Operation of the Third Embodiment Next, the operation of the moving image distribution system B of the third embodiment having the above configuration will be described.

  Hereinafter, of the operation of the moving image distribution system B according to the third embodiment, differences from the first embodiment will be described with reference to FIGS. 12 and 13.

  The moving image distribution system B according to the third embodiment is different in that the above-described steps S106, S112, and S113 are replaced with steps S306, S312 and S313, respectively.

  First, the process (step S306) on the moving image encoding device 10B side will be described.

  As shown in FIG. 12, step S306 is divided into three steps S306-1 to S306-3.

  In step S306-1, the rate control unit 107B specifies a position other than the padding bits based on the bit depth control signal S12.

  In step S306-2, the rate control unit 107B obtains a bit inversion probability for only bits other than the padding bits.

  In step S306-3, the rate R is generated from the obtained bit inversion probability by the rate control unit 107B.

  In the Slepian-Wolf encoding process of the first embodiment, encoding is performed on the assumption that the bit inversion probability is steady, so that the Sleiam-Wolf code in consideration of the possibility of bit inversion occurring at the position of the padding bit. Is generated. However, the position of the padding bit is an area where information (bit depth control signal S12) is shared between the encoding side and the decoding side, and no new information is required. Therefore, in the video decoding device 20B of the third embodiment, the position of the padding bit is specified, and the bit inversion probability and the rate R are calculated using only the bits other than the padding bit, thereby preventing an increase in the code amount. .

  Next, the process (steps S312 and S313) on the video decoding device 20B side will be described.

  As shown in FIG. 13, step S312 is divided into two steps S312-1 and S312-2.

  In step S312-1, the LLR estimation unit 203 estimates the LLR from the quantized decoder predicted image F32 and outputs the LLR notification signal S22.

  In step S312-2, the LLR correction unit 209 corrects the LLR notification signal S22 based on the bit depth control signal S11, and generates the corrected LLR notification signal S23.

  Specifically, the LLR correction unit 209 recognizes the position of the padding bit from the bit depth control signal S11, and the size of the LLR corresponding to the bit other than the padding bit is sufficiently larger than the LLR of the padding bit. Correct as follows.

  In step S313, the Slepian-Wol decoding unit 204 performs decoding using the corrected LLR notification signal S23 and the Slepian-Wolf stream ST1, and generates a corrected Wyner-Ziv image F33.

  In the Slepian-Wolf decoding process of the second embodiment, since the process is performed on the assumption that the bit inversion probability is steady, the Sleiam-Wolf decoding process considering the possibility of bit inversion occurring even at the padding bit position is performed. Do. However, the position of the padding bit is an area where information (bit depth control signal S12) is shared between the encoding side and the decoding side, and no new information is required. Therefore, in the video decoding device 20B of the third embodiment, it is possible to improve the decoding success probability (decoding quality) by specifying the position of the padding bit and correcting the LLR. In particular, as in the moving image encoding apparatus 10B of the third embodiment, the position of padding bits is specified, the bit inversion probability and the rate R are calculated only with bits other than the padding bits, and the code amount is reduced. If so, the effect of improving the decoding quality by correcting the LLR is increased.

(C-3) Effects of the Third Embodiment According to the third embodiment, the following effects can be achieved as compared with the second embodiment.

  In the video encoding device 10B, the rate is estimated only for bits other than the padding bits for which information is not shared with the video decoding device 20B, thereby reducing the amount of codes and efficient coding. Can be made.

  Also, in the video decoding device 20B, the information sharing with the video encoding device 10B is facilitated by increasing the LLR of the padding bits larger than the bits other than the padding bits. And decoding quality can be improved.

(D) Fourth Embodiment Hereinafter, a fourth embodiment of a moving image encoding device and program, a moving image decoding device and program, and a moving image distribution system according to the present invention will be described in detail with reference to the drawings. .

(D-1) Configuration of Fourth Embodiment The overall configuration of a moving image distribution system 1C of the fourth embodiment can also be shown using FIG. Hereinafter, the difference between the fourth embodiment and the second embodiment will be described.

  The moving image distribution system 1C of the fourth embodiment is different from the second embodiment in that the moving image encoding device 10A and the moving image decoding device 20 can be replaced with the moving image encoding device 10C.

  FIG. 14 is a block diagram illustrating a functional configuration of a moving image encoding device 10C according to the fourth embodiment.

  In the moving picture encoding apparatus 10C of the fourth embodiment, the second is that the Slepian-Wolf encoding unit 108 is replaced by the Lepian-Wolf encoding unit 108C, and a Slepian-Wolf code selection unit 110 is added. This is different from the embodiment.

  The Slepian-Wolf code selection unit 110 identifies bits other than padding bits based on the bit depth control signal S12, and information for performing the Slepian-Wolf encoding process on the identified bits (hereinafter referred to as “dedicated Slepian”). -Wolf encoded information S13 ").

  In this embodiment, the dedicated Slepian-Wolf encoding information S13 is an optimal Slepian-Wolf code (Slepian-Wolf code) that is optimal for encoding with the number of bits of the above specified bits (the number of bits to be encoded). Code matrix used for conversion).

  For example, the Slepian-Wolf code selection unit 110 stores in advance an optimum Lepian-Wolf code for each number of bits to be encoded (specific number of bits), and the number of bits to be encoded (specific number of bits). It is also possible to select a Slepian-Wolf code according to the above and use it for generating the dedicated Slepian-Wolf encoded information S13.

  The specific process of holding the Slepian-Wolf code corresponding to an arbitrary number of bits in the Slepian-Wolf code selection unit 110 is not limited. For example, an LDPC code generated using the technique described in Reference 1 (Xiao-Yu Hu, Regular and irregular progressive edge-growth tanner graphs, Information Theory, IEEE Transactions on, Volume 51, Issue 1, Pages 386-398.) (Low-Density Parity-check Code), Reference 2 (D. Varodayan, A. Aaron and B. Girod, Rate-adaptive codes for distributed source coding, EURASIP Signal Processing Journal, Special Section on Distributed Source Coding, vol. 86, no. 11, pp. 3123-3130, November 2006.), an LDPC code whose rate can be easily changed is obtained. Therefore, the Slepian-Wolf code selection unit 110 holds, for example, the above-described LDPC code adjusted according to an arbitrary number of bits as the Slepian-Wolf code using the techniques described in the above-mentioned References 1 and 2. You may do it.

  The Slepian-Wolf encoding unit 108C performs Slepian-Wolf encoding using the dedicated Slepian-Wolf encoding information S13, and generates a Slepian-Wolf stream ST1.

  FIG. 15 is a block diagram illustrating a functional configuration of a video decoding device 20C according to the fourth embodiment.

  In the moving picture decoding apparatus 20C according to the fourth embodiment, the second embodiment is that a Slepian-Wol decoding unit 204 is replaced with a Lepian-Wol decoding unit 204C, and a Slepian-Wolf code selection unit 208 is further added. Is different.

  The Slepian-Wolf code selection unit 208 specifies bits other than the padding bits based on the bit depth control signal S11, and performs information on the specified bit for the Slepian-Wolf decoding process (hereinafter referred to as “dedicated Slepian- Wolf decoding information S24 ").

  The dedicated Slepian-Wolf decoding information S24 includes a Slepian-Wolf code (a matrix of codes used for Slepian-Wolf decoding) that is optimal for decoding with the number of bits specified above.

  For example, the Slepian-Wolf code selection unit 208 holds in advance an optimal Lepian-Wolf code for each bit number (specific bit number) to be decoded, and according to the bit number (specific bit number) to be decoded. Alternatively, the selected Lepian-Wolf code may be selected and used to generate the dedicated Lepian-Wolf decoding information S24. The Slepian-Wolf code selection unit 208 holds, for example, the dedicated Slepian-Wolf encoded information S13 generated by using the techniques described in the above-mentioned References 1 and 2 in the same manner as the above-mentioned Lepian-Wolf code selection unit 110. You may make it do.

  The Slepian-Wolf encoding unit 108C is different from the second embodiment in that it performs Slepian-Wolf decoding using the dedicated Slepian-Wolf encoding information S13 and generates a corrected Wyner-Ziv image F33.

(C-2) Operation of the Fourth Embodiment Next, the operation of the moving image distribution system 1C of the fourth embodiment having the above configuration will be described.

  Hereinafter, of the operation of the moving image distribution system 1C according to the fourth embodiment, differences from the second embodiment will be described with reference to FIGS.

  The moving image distribution system 1A according to the fourth embodiment is different from the second embodiment in that steps S107 and S113 described above are replaced with steps S407 and S413, respectively.

  First, the process (step S407) on the moving image encoding apparatus 10C side will be described.

  As shown in FIG. 16, step S407 is divided into two steps S407-1 and S407-2.

  In step S407-1, the Slepian-Wolf code selection unit 110 specifies bits other than padding bits based on the bit depth control signal S12, and dedicated Slepian-Wolf encoded information S13 is generated based on the specified bits. The

  In step S407-2, the quantized Wyner-Ziv image F23 is encoded by the Slepian-Wolf encoding unit 108C based on the rate R, using the dedicated Slepian-Wolf encoding information S13, and the Slepian-Wolf stream ST1. Is generated. Specifically, the Slepian-Wolf encoding unit 108C collects bits to be encoded based on the information of the dedicated Slepian-Wolf encoding information S13, and encodes the collected bits with the above-described optimal Sleian-Wolf code. .

  Next, the process (step S413) on the moving image decoding apparatus 20A side will be described.

  As shown in FIG. 17, step S413 is divided into two steps S413-1 and S413-2.

  In step S413-1, the Slepian-Wolf code selection unit 208 specifies bits other than padding bits based on the bit depth control signal S11, and generates dedicated Slepian-Wolf decoding information S24 based on the specified bits. .

  In step S413-2, the Slepian-Wol decoding unit 204C uses the LLR notification signal S22 and the dedicated Slepian-Wolf decoding information S24 to decode the Slepian-Wolf stream ST1, and generates a corrected Wyner-Ziv image F33. The

  Specifically, the Slepian-Wol decoding unit 204C collects LLRs corresponding to the bits to be decoded based on the information of the dedicated Slepian-Wolf decoding information S24, and the collected LLRs are converted into the above-mentioned optimal Sleian-Wolf codes using the above-described Slepian-Wolf codes. -Decodes the Wolf stream ST1.

(D-3) Effects of Fourth Embodiment According to the fourth embodiment, the following effects can be obtained in addition to the effects of the second embodiment.

  The moving image encoding apparatus 10C uses the Slipian-Wolf code that is optimal for encoding bits other than the padding bits by using the fact that the position of padding bits is shared with the moving image decoding apparatus 20C. Thus, an increase in code amount can be prevented.

  Also, the moving picture decoding apparatus 20C specifies the position of padding bits based on the bit depth control signal S12 supplied from the moving picture coding apparatus 10C, and is optimal for correcting errors of bits other than the padding bits. Decoding quality can be improved by decoding using a code.

(E) Other Embodiments The present invention is not limited to the above-described embodiments, and may include modified embodiments as exemplified below.

(F-1) In each of the above embodiments, for simplification of description, for each image (Wyner-Ziv image, encoder predicted image, decoder predicted image, etc.) used for encoding and decoding a Wyner-Ziv image, As described in Non-Patent Document 1, etc., it is assumed that processing is performed in the parameter format (pixel domain format) represented by the pixel value or quantized value of each pixel without performing conversion processing by DCT conversion or the like. However, in the present invention, the parameter format for representing each image is not limited to the pixel domain format, and may be a parameter format (transform domain format) represented by a quantized value for each transform coefficient region by DCT transformation or the like. good. In the case of the transform domain format, the pixel value (quantized value) of the pixel in each of the above embodiments is a transform coefficient region (in the case of processing divided into a plurality of blocks, the transform coefficient region in each block). It will be replaced with the quantized value of each. In each of the above embodiments, when each image used for encoding and decoding is processed in the transform domain format, a conversion processing unit that converts the parameter sequence in the transform domain format is added before the quantization process. Just do it.

  In each of the above embodiments, the pixel value of each pixel can be viewed as a parameter string arranged as a parameter. In the DVC method encoding process and decoding process based on the Slepian-Wolf theory and the Wyner-Ziv theory, the number of parameter sequences to be encoded and decoded and the meaning of each parameter are not limited. Therefore, even if the format of the parameter string for expressing each image (number of parameters, etc.) is changed, the same effects as those of the above embodiments can be obtained.

(F-2) In each of the above embodiments, the moving image decoding apparatus generates an encoder predicted image and a decoder predicted image from a Key image. However, the moving image decoding device may generate a Wyner-Ziv decoded image at another time. Moreover, in a moving image decoding apparatus, when it has a scalable structure like the nonpatent literature 3, you may produce | generate a decoder prediction image using the information of a base layer.

(F-3) In the video encoding device of each of the embodiments described above, the rate R is determined based on the estimation result of the rate control unit. The rate control may be performed by a feedback approach from the side. When rate control is performed by a feedback approach, the encoder predicted image quantization unit 106 may be omitted.

(F-4) In each of the above embodiments, the video encoding device and the video decoding device are described as being connected via a communication path such as a network. However, the video encoding device and the video decoding are described. The apparatus may not be configured to be able to communicate directly. For example, the moving image data (Slepian-Wolf stream and Key stream data) generated by the moving image encoding device may be recorded on a medium such as a DVD or a hard disk and supplied to the moving image decoding device offline. .

(F-5) In the fourth embodiment, as a form of the Slepian-Wolf code selection unit 110 and the Slepian-Wolf code selection unit 208, the bits belonging to the pixel for which the encoder predicted image F21 is selected are encoded and decoded. A Slepian-Wolf code that is not related to processing may be generated. In the Slepian-Wolf code, the position information of the bits to be encoded and decoded is inherent in the Slepian-Wolf code. In this case, the moving image distribution system 1C does not need to collect the bits to be encoded in step S407-2 or the LLRs corresponding to the bits to be decoded in step S313-2.

(F-6) In each of the above embodiments, the bit depth control unit determines the bit depth for each block, but determines the bit depth for each pixel (for each parameter) and reflects it in the bit depth control signal S11. You may make it do.

  DESCRIPTION OF SYMBOLS 1 ... Moving image delivery system, 10 ... Moving image encoding apparatus, 101 ... Bit depth control part, 102 ... Complexity calculation part, 103 ... Control part 103, 104 ... Encoding object image quantization part, 105 ... Encoder prediction image Generation unit 106 ... Encoder prediction image quantization unit 107 ... Rate control unit 108 ... Slepian-Wolf encoding unit 109 ... Bit depth notification unit 120 ... Key image encoding unit 20 ... Video decoding device 200 DESCRIPTION OF SYMBOLS ... Bit depth receiving part, 201 ... Decoder prediction image generation part, 202 ... Decoder prediction image quantization part, 203 ... LLR estimation part, 204 ... Slepian-Wol decoding part, 205 ... Reconstruction part, 220 ... Key image decoding part.

Claims (11)

In a video encoding device that encodes a video having a frame sequence,
Predicted image generation means for generating a predicted image of a non-key frame using a key frame of the frame sequence;
A complexity calculator for calculating the complexity of non-keyframes;
Based on the calculation result of the complexity calculation unit, bit depth determination means for determining a bit depth related to quantization of each parameter constituting the image;
Quantize each parameter of the non-key frame and each parameter of the predicted image with the bit depth determined by the bit depth determining means to generate a non-key frame after quantization and a predicted image after quantization. For a parameter for which a bit depth less than the maximum bit depth is determined, either the most significant bit side or the least significant bit side of the quantized data corresponding to the bit depth determined for the parameter, Quantization means for generating, as quantized data of the parameter, data obtained by adding padding bits of a predetermined pattern corresponding to the difference from the most bit depth,
An error correction code generation means for generating an error correction code for correcting an error of the predicted image after quantization for the non-key frame after quantization;
Parameters of each image processed by the bit depth determination means and the quantizing means, the pixel values of the pixels constituting the image, or, conversion coefficients obtained by converting the image by a predetermined conversion method A moving picture coding apparatus characterized by having a quantized value for each region.   The apparatus further comprises output means for outputting data including the error correction code generated by the error correction code generation means and the bit depth information related to the bit depth determined by the bit depth determination means as data relating to a non-key frame. The moving picture encoding apparatus according to claim 1. The complexity calculation unit calculates the complexity for each block constituting the predicted image of the non-key frame,
The moving picture encoding apparatus according to claim 1 or 2, wherein the bit depth determining means determines a common bit depth for each parameter in the block for each block constituting the image.   The output rate of the error correction code output by the error correction code generation means is determined, and an output rate necessary for error correction processing only for bits other than padding bits in the non-key frame after quantization is determined. 4. The moving picture encoding apparatus according to claim 3, further comprising rate control means for determining.   The moving image according to any one of claims 1 to 4, wherein the error correction code generation means generates an error correction code only for bits other than padding bits in a non-key frame after quantization. Encoding device. In a moving image decoding apparatus for decoding moving image data in which a moving image having a frame sequence is encoded in units of frames,
Key frame holding means for holding key frame data;
A predicted image generating means for generating a predicted image of a non-key frame in the frame sequence using the key frame held by the key frame holding means;
Bit depth holding means for holding bit depth information related to bit depth applied to quantization of each parameter constituting the image in the process of generating encoded data of non-key frames constituting the moving image data;
The bit depth held by the bit depth holding unit quantizes each parameter of the predicted image to generate a quantized predicted image, and a bit depth smaller than the maximum bit depth is determined. For a parameter, padding bits of a predetermined pattern corresponding to the difference from the most bit depth are placed on either the most significant bit side or the least significant bit side of the quantized data for the bit depth set for the parameter. Quantization means for generating the added data as quantized data of the parameter;
When the encoded data of the non-key frame that constitutes the moving image data is an error correction code for correcting the predicted image after quantization, the corrected image obtained by correcting the predicted image after quantization using the error correction code Error correction means for generating an image;
Reconstructing means for generating a reconstructed image obtained by reconstructing a non-keyframe image based on the corrected image and the predicted image;
Parameters of each image processed by the bit depth holding means and the quantizing means, the pixel values of the pixels constituting the image, or, conversion coefficients obtained by converting the image by a predetermined conversion method A moving picture decoding apparatus characterized by having a quantized value for each region. The error correction means is
An LLR determining unit that determines an LLR for each bit constituting the quantized predicted image;
An error correction process is performed based on the LLR for each bit determined by the LLR determination unit,
The moving picture decoding apparatus according to claim 6, wherein the LLR determination unit sets an LLR having a larger value for padding bits than a bit excluding padding bits.   8. The moving picture decoding apparatus according to claim 6 or 7, wherein the error correction means performs error correction only for bits excluding padding bits in the predicted image after quantization. A computer mounted on a video encoding device that encodes a video having a frame sequence,
Predicted image generation means for generating a predicted image of a non-key frame using a key frame of the frame sequence;
A complexity calculator for calculating the complexity of non-keyframes;
Based on the calculation result of the complexity calculation unit, bit depth determination means for determining a bit depth related to quantization of each parameter constituting the image;
Quantize each parameter of the non-key frame and each parameter of the predicted image with the bit depth determined by the bit depth determining means to generate a non-key frame after quantization and a predicted image after quantization. For a parameter for which a bit depth less than the maximum bit depth is determined, either the most significant bit side or the least significant bit side of the quantized data corresponding to the bit depth determined for the parameter, Quantization means for generating, as quantized data of the parameter, data obtained by adding padding bits of a predetermined pattern corresponding to the difference from the most bit depth,
Function as an error correction code generation means for generating an error correction code for correcting an error of a predicted image after quantization for a non-key frame after quantization;
Parameters of each image processed by the bit depth determination means and the quantizing means, the pixel values of the pixels constituting the image, or, conversion coefficients obtained by converting the image by a predetermined conversion method A moving picture coding program characterized by being a quantized value for each region. A computer mounted on a moving image decoding apparatus for decoding moving image data in which a moving image having a frame sequence is encoded in units of frames;
Key frame holding means for holding key frame data;
A predicted image generating means for generating a predicted image of a non-key frame in the frame sequence using the key frame held by the key frame holding means;
Bit depth holding means for holding bit depth information related to bit depth applied to quantization of each parameter constituting the image in the process of generating encoded data of non-key frames constituting the moving image data;
The bit depth held by the bit depth holding unit quantizes each parameter of the predicted image to generate a quantized predicted image, and a bit depth smaller than the maximum bit depth is determined. For a parameter, padding bits of a predetermined pattern corresponding to the difference from the most bit depth are placed on either the most significant bit side or the least significant bit side of the quantized data for the bit depth set for the parameter. Quantization means for generating the added data as quantized data of the parameter;
When the encoded data of the non-key frame that constitutes the moving image data is an error correction code for correcting the predicted image after quantization, the corrected image obtained by correcting the predicted image after quantization using the error correction code Error correction means for generating an image;
Based on the corrected image and the predicted image, function as a reconstruction unit that generates a reconstructed image obtained by reconstructing a non-keyframe image;
Parameters of each image processed by the bit depth holding means and the quantizing means, the pixel values of the pixels constituting the image, or, conversion coefficients obtained by converting the image by a predetermined conversion method A moving picture decoding program characterized by being a quantized value for each area. A moving image including a moving image encoding device that generates moving image data obtained by encoding a moving image having a frame sequence in units of frames, and a moving image decoding device that decodes the moving image data generated by the moving image encoding device. In the image distribution system,
Applying the video encoding device according to claim 1 as the video encoding device,
A moving image distribution system, wherein the moving image decoding device according to claim 6 is applied as the moving image decoding device.

Images