JP2008227670A - Image coding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image coding device with reduced power consumption, capable of effectively narrowing down the number of prediction modes and reducing the operation amount at coding and realizes coding of high coding efficiency. <P>SOLUTION: The device comprises a coding means 112 for selecting one prediction mode from a plurality of prediction modes and performing intra prediction coding; a prediction error detecting means 113 for detecting a prediction error; an operation setting means 132 for setting up a standard operation method and a high-definition operation method; and a mode setting means 111 for setting up first candidate modes and second candidate modes. In the standard operation, an intra-control portion 13a controls the coding means 112 so as to select a first candidate mode among the first candidate modes as the prediction mode, when a prediction error in the selected first candidate mode is lower than the threshold, and select the prediction mode from all prediction modes, when there is no prediction mode below the threshold. In the high-definition operation, upon detection of a prediction error for the first candidate modes, the intra control portion 13a controls the coding means 112 so as to select the prediction mode of a prediction error lower than a big expanded threshold or select the prediction mode with a minimum prediction error among the second candidate modes. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention selects an input image signal from a plurality of prepared intra prediction modes and performs intra prediction encoding, and performs inter prediction encoding by motion compensation to generate an encoded image signal. The present invention relates to an image encoding device that obtains.

Recently, H.H. is one of the video compression standards. H.264 / MPEG (moving picture experts group) -4 Video phone using AVC (Advanced Video Coding), terrestrial digital broadcasting called so-called 1-seg system, and moving image recording apparatus such as movies have come into practical use. I came.
On the other hand, the portable moving image recording apparatus is a small device that operates using a battery as a power source, and it is preferable that a small and power-saving recording apparatus can be realized.

Patent Document 1 describes H.264. An image processing apparatus that uses the H.264 / MPEG-4 AVC encoding method and can perform intra prediction encoding in a short time with a small-scale configuration is disclosed. That is, the apparatus includes a difference detection circuit that detects a difference sum in the horizontal and vertical directions of pixel data in a macroblock, and an intra that matches the correlation direction by specifying the correlation direction of the macroblock using the detected difference sum. A small-scale image processing apparatus is realized by including a mode selection circuit for selecting a prediction mode and an encoding circuit for performing intra prediction encoding in the selected intra prediction mode.
JP 2004-304724 A

  However, the image processing device disclosed in Patent Document 1 simply selects a prediction mode by detecting a horizontal correlation value or a vertical correlation value of a pixel block luminance level to be encoded, The number of prediction modes cannot be narrowed down sufficiently, reducing the amount of computation during coding, reducing the power required for computation, and realizing an image processing device that performs coding with high coding efficiency I couldn't.

  Therefore, the present invention has been made to solve the above-described problems, effectively reducing the number of prediction modes, reducing the amount of calculation during encoding, and reducing power consumption. An object of the present invention is to provide an image coding apparatus capable of performing coding with high coding efficiency.

According to a first aspect of the present invention, there is provided an image encoding apparatus that divides an input image into a predetermined number of vertical and horizontal pixels, and encodes a block image obtained by the division while sequentially performing intra prediction. Mode group setting means for setting two or more prediction modes selected in advance as a first candidate mode group, adding another prediction mode to the first candidate mode group, and setting as a second candidate mode group The operation setting means for detecting the number of pixels of the input image and setting the operation to be standard when the number of pixels is within a predetermined number and to be high-definition when exceeding the predetermined number; and the operation setting means When set as the standard, the predicted image obtained by intra-predicting the block image in each prediction mode of the first candidate mode group set by the mode group setting means and the block A prediction error that is a difference from an image is obtained as a prediction error corresponding to each prediction mode, and when a prediction error equal to or less than a predetermined threshold exists in each of the obtained prediction errors, a prediction mode that gives the prediction error is set to 1 On the other hand, when there is no prediction error equal to or less than the predetermined threshold value in each of the obtained prediction errors, the respective prediction errors corresponding to all of the plurality of prediction modes are obtained, and each of the obtained prediction errors is selected. Each of the first candidate mode group set by the mode group setting unit when the prediction mode that gives the smallest prediction error is selected and set as the high definition by the operation setting unit Prediction error corresponding to each prediction mode, which is a difference between a prediction image obtained by intra prediction of the block image in the mode and the block image When there is a prediction error equal to or less than the enlargement threshold value that is larger than the predetermined threshold value in each of the obtained prediction errors, one prediction mode that gives the prediction error is selected, When there is no prediction error equal to or smaller than the expansion threshold in the prediction error, the prediction error corresponding to the prediction mode of the second candidate mode group is obtained, and the smallest prediction error among the obtained prediction errors And a mode selection unit that selects one prediction mode that provides the image, and an encoding unit that performs intra-prediction coding on the block image using one prediction mode selected by the mode selection unit to obtain a coded signal. An image encoding device characterized by the above is provided.
The second invention divides an input image into a predetermined number of pixels in the vertical and horizontal directions, and encodes a block image obtained by the division while sequentially performing intra prediction and motion prediction according to a predetermined rule. In the image encoding device that generates the image, a mode group setting unit that sets two or more preselected prediction modes as a candidate mode group among a plurality of prediction modes, and an input image is encoded and decoded Search range setting in which the obtained decoded image is used as a reference image, and a search range for searching for a motion vector by searching for a matching position of the block image on the reference image is set as a standard range and an enlarged range larger than the standard range And a motion prediction immediately before the input image with respect to a code amount of an encoded signal generated by intra prediction immediately before the input image. A code amount ratio detecting unit that detects a code amount ratio of the encoded signal as a code amount ratio; and when the code amount ratio detected by the code amount ratio detecting unit is larger than a predetermined threshold, Prediction in which the block image is intra-predicted in each prediction mode of the candidate mode group set by the mode group setting means, and a difference between each prediction image obtained by intra prediction and the block image is associated with each prediction mode The prediction mode which is obtained as an error and gives the smallest prediction error among the obtained prediction errors is selected as the prediction mode for the intra prediction, and the enlarged range set by the search range setting means is selected as the motion When the code amount ratio detected by the code amount ratio detecting means is selected as the search range of the motion vector to be obtained in the prediction, the code amount ratio is not more than the predetermined threshold value. The intra prediction of the block image in all of the plurality of prediction modes, and the difference between each prediction image obtained by intra prediction and the block image is obtained as a prediction error corresponding to each prediction mode. The prediction mode that gives the smallest prediction error among the prediction errors is selected as the prediction mode for the intra prediction, and the standard range selected by the search range setting means is obtained when the motion prediction is performed. A selection unit that selects a motion vector search range; an intra encoding unit that obtains an intra prediction encoded signal by intra prediction encoding the block image using the prediction mode selected by the selection unit; and the selection unit. A motion prediction coding signal is obtained by performing motion prediction coding using a motion vector obtained by the selected search range. A motion prediction encoding unit, an intra prediction encoded signal obtained by the intra encoding unit, and a motion prediction encoded signal obtained by the motion prediction encoding unit are entropy encoded to generate the encoded signal. And an entropy encoding means.

According to the present invention, two or more prediction modes selected in advance among a plurality of prediction modes are set as a first candidate mode group, and another prediction mode is added to the first candidate mode group to obtain a second candidate mode group. A mode group setting means for setting the number of pixels of the input image, and an operation setting means for setting a standard operation when the number of pixels is within a predetermined number and setting a high-definition operation when the number exceeds the predetermined number; The difference between the predicted image and the block image obtained by intra-predicting the block image in each prediction mode of the first candidate mode group set by the mode group setting unit when set as the standard by the operation setting unit The prediction error is obtained as a prediction error corresponding to each prediction mode, and when a prediction error equal to or less than a predetermined threshold exists in each of the obtained prediction errors, one prediction mode that gives the prediction error is selected. On the other hand, if there is no prediction error equal to or less than the predetermined threshold in each of the obtained prediction errors, the respective prediction errors corresponding to all of the plurality of prediction modes are obtained, and the largest of the obtained prediction errors. When one prediction mode giving a small prediction error is selected and set as high definition by the operation setting unit, the block image is intra predicted in each prediction mode of the first candidate mode group set by the mode group setting unit. The prediction error, which is the difference between the prediction image obtained in this way and the block image, is obtained as a prediction error corresponding to each prediction mode, and the prediction error below the enlargement threshold value that is larger than the predetermined threshold value for each of the obtained prediction errors If there is a prediction error, select one prediction mode that gives the prediction error, while each calculated prediction error has a prediction error less than the expansion threshold If not, mode selection means for obtaining each prediction error corresponding to the prediction mode of the second candidate mode group and selecting one prediction mode that gives the smallest prediction error among the obtained prediction errors; Since there is a special configuration including an encoding unit that obtains an encoded signal by intra prediction encoding a block image using one prediction mode selected by the mode selection unit, the number of prediction modes is effectively narrowed down Thus, it is possible to realize an image coding apparatus that can reduce power consumption, such as reducing the amount of calculation during coding, and perform coding with high coding efficiency.
Also, mode group setting means for setting two or more pre-selected prediction modes among a plurality of prediction modes as candidate mode groups, and a decoded image obtained by encoding and decoding an input image as a reference image Search range setting means for setting a search range when searching for a motion vector by searching for a matching position of a block image on a reference image as a standard range and an enlarged range larger than the standard range, and immediately before the input image A code amount ratio detecting means for detecting, as a code amount ratio, a ratio of a code amount of an encoded signal generated by motion prediction immediately before an input image to a code amount of an encoded signal generated by intra prediction; When the code amount ratio detected by the code amount ratio detecting unit is larger than a predetermined threshold, the block image is loaded in each prediction mode of the candidate mode group set by the mode group setting unit. Prediction mode that gives the smallest prediction error among the obtained prediction errors by calculating the difference between each prediction image obtained by intra prediction and intra prediction as a prediction error corresponding to each prediction mode Is selected as a prediction mode for intra prediction, and an enlarged range set by the search range setting means is selected as a motion vector search range obtained when motion prediction is performed, while a code detected by the code amount ratio detection means is selected. When the quantity ratio is equal to or less than a predetermined threshold, the block image is intra-predicted in all of the plurality of prediction modes, and the difference between each prediction image obtained by intra prediction and the block image is associated with each prediction mode. As a prediction error, the prediction mode that gives the smallest prediction error among the calculated prediction errors is the prediction mode for intra prediction. A selection unit that selects the standard range selected by the search range setting unit as a search range of a motion vector obtained when motion prediction is performed, and an intra prediction code for the block image using the prediction mode selected by the selection unit And intra-coding means for obtaining an intra-predictive coded signal, and motion-predictive coding means for obtaining a motion-predicted coded signal by performing motion-predictive coding using a motion vector obtained from the search range selected by the selecting means; In the case of comprising entropy encoding means for entropy encoding the intra prediction encoded signal obtained by the intra encoding means and the motion prediction encoded signal obtained by the motion prediction encoding means to generate an encoded signal. Effectively narrow down the number of prediction modes and search for motion vectors, reduce the amount of calculation at the time of encoding, and reduce power consumption, and An image coding apparatus capable of performing coding with high coding efficiency can be realized.

Hereinafter, an image encoding apparatus according to an embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a block diagram illustrating a configuration example of an image encoding device according to an embodiment of the present invention. FIG. 2 is a diagram showing a configuration example (No. 1) of a main part of the image coding apparatus according to the embodiment of the present invention. FIG. 3 is a diagram illustrating a configuration example (No. 2) of a main part of the image encoding device according to the embodiment of the present invention. FIG. 4 is a flowchart showing the operation of the image coding apparatus according to the embodiment of the present invention. FIG. 5 is a diagram for describing each mode of 16 × 16 pixel block intra prediction. FIG. 6 is a diagram for describing each mode of intra prediction of a 4 × 4 pixel block. FIG. 7 is a diagram (part 1) illustrating the intra prediction operation according to the embodiment of the present invention in a flowchart. FIG. 8 is a flowchart (part 2) illustrating the intra prediction operation according to the embodiment of the present invention. FIG. 9 is a diagram (part 3) illustrating the intra prediction operation according to the embodiment of the present invention in a flowchart. FIG. 10 is a diagram (part 4) illustrating the intra prediction operation according to the embodiment of the present invention in a flowchart.

  The image encoding apparatus effectively reduces the number of prediction modes, reduces the amount of calculation at the time of encoding, reduces power consumption, and can perform encoding with high encoding efficiency The second candidate mode is obtained by adding two or more prediction modes selected in advance as a first candidate mode group and adding another prediction mode to the first candidate mode group. Mode group setting means for setting as a group, and an operation setting means for detecting the number of pixels of an input image and setting an operation to be standard when the number of pixels is within a predetermined number and to be high-definition when exceeding the predetermined number And the difference between the prediction image obtained by intra-predicting the block image in each prediction mode of the first candidate mode group set by the mode group setting unit and the block image when set as a standard by the operation setting unit A prediction error is obtained as a prediction error corresponding to each prediction mode, and when a prediction error equal to or less than a predetermined threshold exists in each of the obtained prediction errors, one prediction mode that gives the prediction error is selected. When there is no prediction error equal to or less than the predetermined threshold in each of the obtained prediction errors, the respective prediction errors corresponding to all the plurality of prediction modes are obtained, and the smallest prediction error among the obtained prediction errors Is obtained by intra-predicting a block image in each prediction mode of the first candidate mode group set by the mode group setting means when one prediction mode is given and set as high definition by the operation setting means A prediction error that is a difference between the prediction image and the block image is obtained as a prediction error corresponding to each prediction mode, and a predetermined value is set for each of the obtained prediction errors. If there is a prediction error that is greater than the threshold value that is less than or equal to the enlargement threshold value, one prediction mode that gives the prediction error is selected, while there is no prediction error that is less than or equal to the enlargement threshold value for each of the obtained prediction errors Mode selection means for obtaining each prediction error corresponding to the prediction mode of the second candidate mode group and selecting one prediction mode that gives the smallest prediction error among the obtained prediction errors; and mode selection means And an encoding unit that obtains an encoded signal by intra-predicting the block image using one prediction mode selected by the above.

The configuration of the image encoding device will be described.
An image encoding apparatus 1 illustrated in FIG. 1 includes an intra prediction unit 11, a subtractor 12, an encoding control unit 13, a conversion unit 14, a quantization unit 15, an entropy encoding unit 16, an inverse quantization unit 17, and an inverse conversion unit. 18, an adder 19, a loop filter 21, a frame memory 22, a motion compensation unit 23, a motion detection unit 24, and a switch 25. The encoding control unit 13 includes an intra control unit 13a and an inter control unit 13b.
The intra control unit 13 a illustrated in FIG. 2 includes a pixel number detector 131, a pixel number determination unit 132, and an intra controller 133. The intra prediction unit 11 includes a prediction mode setting unit 111, an intra prediction executor 112, a prediction error calculator 113, and an encoded data acquisition unit 114.
The inter control unit 13b illustrated in FIG. 3 includes a code amount acquisition unit 135, an inter code amount storage unit 136, an intra code amount storage unit 137, a code amount comparator 138, and a condition determination unit 139. The motion detection unit 24 includes a search range setting unit 241, a search accuracy setting unit 242, and a motion detection execution unit 243.

The operation of the image coding apparatus will be described.
First, the intra control unit 13a of the encoding control unit 13 detects whether the input image signal is a standard image or a high-definition image. When the input image signal is a standard image, the intra control unit 13a generates an evaluation threshold for causing the calculation accuracy of the intra prediction performed by the intra prediction unit 11 to be performed with standard accuracy, and the input image signal is a high-definition image. In this case, an evaluation threshold value that has a calculation accuracy larger than the standard accuracy is generated. The intra prediction unit 11 converts an input image signal into, for example, H.264. H.264 / MPEG-4 AVC encoding system is used for intra prediction encoding. The encoding is performed when the macroblock image divided into 16 × 16 pixels is encoded according to the four prediction modes, and when the macroblock image is divided into 16 4 × 4 pixel blocks, the respective blocks are divided. In some cases, encoding is performed according to each of nine prediction modes. The intra control unit 13a obtains a prediction evaluation value in each combination, and generates an encoded signal that is encoded using a mode that gives the smallest prediction evaluation value based on the prediction evaluation value. If a mode that gives the smallest prediction evaluation value is selected, an encoded signal with the smallest code amount can be obtained.

  In order to obtain predictive evaluation values for all combinations of modes for 16 4 × 4 pixel blocks, the amount of computation increases and the power consumption also increases. Therefore, a mode that gives a small prediction evaluation value is predicted in advance, and a calculation of a prediction evaluation value for a mode that gives a large prediction evaluation value is omitted. That is, the correlation between the pixels constituting the block is examined to determine whether the correlation between the pixels is in the horizontal direction, the vertical direction, the right-down direction, or the right-up direction, and similar to those directions. The selected direction is selected as a prediction direction. Predictive evaluation values are sequentially obtained for the selected prediction mode candidates. Further, a small evaluation threshold is set, and when a predicted evaluation value smaller than the set evaluation threshold is obtained, the calculation of the predicted evaluation value is stopped there. Intra prediction calculation is performed in a mode in which the prediction evaluation value is obtained. After the calculation is completed, intra prediction calculation of the next macroblock image is performed.

  When a prediction evaluation value having a value smaller than the evaluation threshold cannot be obtained for the selected prediction mode candidate, an intra prediction calculation of the macroblock image is performed in a mode in which the smallest prediction evaluation value is obtained. When the input image signal is a high-definition image, a larger evaluation threshold value is generated than in the case of a standard image. Therefore, the intra prediction calculation for the standard image is performed only when a prediction evaluation value smaller than the small evaluation threshold is obtained, and the calculation for obtaining the prediction evaluation value is interrupted. Is set to a large evaluation threshold, and when a predicted evaluation value smaller than the large threshold is obtained, the calculation for obtaining the predicted evaluation value is interrupted. Compared to the intra prediction operation with a smaller encoding distortion in the case of a standard image, the intra prediction operation includes a little encoding distortion in the case of a high-definition image. The amount of distortion at is small. Calculation time in macroblock units can be shortened. Since the number of macroblocks per frame of a high-definition image is as large as about five times that of a standard image, it is essential to reduce the calculation time for the high-definition image.

Intra prediction encoding and inter prediction encoding performed by the image encoding device 1 will be further described.
First, an intra-predicted image signal obtained by intra-prediction calculation by the intra-prediction unit 11 is supplied as a negative input signal to the subtractor 12 via the terminal a side of the switch 25. The subtracter 12 subtracts the intra predicted image signal from the input image signal that is input to obtain a difference signal. The conversion unit 14 performs DCT (discrete cosine transform) conversion on the difference signal to obtain a coefficient signal. The quantization unit 15 quantizes the coefficient signal with a predetermined quantization width to obtain a quantized signal. The entropy encoding unit 16 entropy encodes the quantized signal and outputs the obtained encoded signal. The intra control unit 13a and the inter control unit 13b of the encoding control unit 13 monitor the code amount of the encoded signal, and when an encoded signal having a code amount larger than the target code amount is obtained, the quantization unit 15 The quantization width is increased and the code amount of the encoded signal is reduced. When the code amount of the encoded signal is smaller than the target code amount, the quantization width is reduced, and control is performed so that an encoded output with a predetermined code amount is obtained.

  The inverse quantization unit 17 inversely quantizes the quantized signal generated by the quantization unit 15 to obtain a decoded coefficient signal. The inverse transform unit 18 performs inverse DCT transform on the decoded coefficient signal to obtain a decoded differential signal. The adder 19 adds the decoded differential signal and the intra predicted image signal obtained by the intra prediction unit 11 to obtain a decoded image. The frame memory 22 stores the decoded image from which the encoding distortion has been removed by the loop filter 21. The motion detection unit 24 obtains a motion vector between an input image signal and a decoded image that is input one frame before and is encoded and decoded. The obtained motion vector is encoded by the entropy encoding unit 16 and output together with the encoded signal output. The motion compensation unit 23 generates a motion compensated image (inter prediction image signal) using the decoded image stored in the frame memory 22 and the motion vector obtained by the motion compensation unit 23. The motion compensated image is supplied to the subtracter 12 and the adder 19 via the terminal b side of the switch 25. The switch 25 is connected to the terminal a side during intra prediction encoding, and is connected to the terminal b side during inter prediction encoding.

  In the image encoding apparatus 1, intra prediction encoding and inter prediction encoding are performed. An image subjected to intra prediction encoding is called an I (Intra-coded) picture. An image subjected to inter-prediction coding is called a P (Predictive-coded) picture in which motion prediction is performed using a past image in time, and an image that is predicted and encoded from the past and future directions. It is called a B (Bidirectionally predictive-coded) picture. The entropy encoding unit 16 generates and outputs an encoded signal together with a header indicating which of the pictures to be encoded according to a predetermined rule is an I, P, or B picture. The inter control unit 13b of the encoding control unit 13 recognizes the code amount of the encoded data of each of the I, P, and B pictures. The amount of code for the input image is small and the ratio of the code amount of the I picture is large when the motion is close to a still image or when the motion is uniform. The ratio of the code amount of P and B pictures increases. For an image in which the ratio of the code amount of an I picture is larger than a predetermined threshold, the encoding accuracy of intra prediction encoding is increased. For example, encoding is performed such that a small evaluation threshold is set by intra encoding and a small prediction evaluation value is obtained. On the other hand, for an image in which the ratio of the code amount of the I picture is smaller than a predetermined threshold, that is, an input image with a large ratio of the code amount of the P and B pictures, for example, the search area in the motion vector calculation is expanded, The accuracy of the motion vector amount is obtained with a precision of 1 pixel or less. Encoding is performed while the prediction error between the input image to be encoded and the prediction image used for motion compensation is kept small.

Next, this will be described in detail.
The pixel number detector 131 of the intra control unit 13a illustrated in FIG. 2 detects the number of pixels of the input image signal. That is, it is detected whether the number of pixels constituting one frame is, for example, a standard image having about 720 × 490 pixels in length and width or a high-definition image having about 1920 × 1080 pixels. The pixel number determination unit 132 determines the standard mode when the input image has a standard image or a pixel number equal to or less than the standard image. In the case of an image having more pixels than the standard image, a determination signal for the high definition mode is generated. The intra controller 133 selects an intra prediction mode candidate for executing an intra prediction calculation with a standard calculation accuracy or higher calculation accuracy than the standard for an image in the standard mode, and obtains an intra prediction. An evaluation threshold value for evaluating the difference value between the predicted image and the input image is set to a small value. In order to increase the calculation speed in the high-definition mode, a smaller number of intra prediction mode candidates than the standard are selected, and a larger threshold is set for the evaluation threshold than the standard mode. Perform the operation.

  The prediction mode setting unit 111 of the intra prediction unit 11 sets a prediction mode for causing the intra controller 133 to execute intra prediction. The prediction mode detects a direction having a high correlation of luminance levels in the horizontal direction, vertical direction, upper right direction, and further lower right direction of the input image divided into blocks. The number of candidates selected by the controller 133 from among the nine intra prediction modes is selected from among the intra prediction mode candidates selected by the intra controller 133. The intra prediction mode from the direction with high correlation is selected. Furthermore, information related to the amount of code at the time of inter prediction coding and the amount of code at the time of intra prediction coding is input to the intra controller 133 from a condition determiner 139 described later, and inter prediction is performed based on the generated control signal. Increase or decrease the number of intra prediction modes used for.

The intra prediction executor 112 performs intra prediction encoding on the input signal in accordance with the set intra prediction mode. For a high-definition mode image, intra prediction encoding is performed with a smaller number of prediction modes than the number of prediction modes in the standard mode. The prediction error calculator 113 compares the predicted image obtained by intra prediction with the intra prediction executor 112 and the input image to obtain a difference value. The difference value is obtained as a luminance level difference between the two images. The evaluation of the difference value is performed by adding the absolute value of the difference value or using the evaluation value obtained by the mean square method. The encoded data acquisition unit 114 sequentially and temporarily stores the prediction images obtained by the intra prediction by the intra prediction execution unit 112, and acquires a prediction image according to the prediction mode designated by the intra controller 133.
The prediction image calculation is performed so that intra prediction calculation of a high-definition image having many macroblocks per frame is completed within one frame period. In the case of a high-definition image, since the size of the macroblock in the screen is relatively small compared to the standard image, the coding distortion is difficult to be visually recognized compared to the distortion of the standard image. In the case of a high-definition image, the amount of power consumed at the time of encoding or the like is larger than that of a standard image. Therefore, it is necessary to reduce the power consumption required at the time of encoding.

The inter prediction calculation will be further described with reference to FIG.
First, the code amount acquisition unit 135 of the inter control unit 13b acquires the code amount information of the I, P, and B pictures encoded from the entropy encoding unit 16 illustrated in FIG. The intra code amount storage device 137 stores the code amount of the I picture, and the inter code amount storage device 136 stores the code amounts of the P and B pictures while tabulating them. The code amount comparator 138 divides the code amount Ri of the encoded output obtained by intra-coding for each GOP (Group of Picture), for example, and the encoded output obtained by the inter-coding by the number of frames, and calculates the average. A code amount Rp is obtained. The ratio of Rp to Ri is calculated as the code amount ratio. The condition determiner 139 sets a predetermined code amount ratio as a threshold T, and if it is determined that the code amount ratio is smaller than the threshold T, the intra prediction coding is performed by adjusting the evaluation threshold set by the intra controller 133 to be small. And a search range setting unit 241 for setting a search range for obtaining a motion vector by the motion detection unit 24, and a setting range for the search range setting unit 242 for setting the search accuracy. A determination signal for increasing the search accuracy. The motion detection executor 243 obtains a motion vector of the input image signal based on the search accuracy of, for example, one pixel set by the search accuracy setting unit 242 in the large reference image region range set by the search range setting unit 241.

When the condition determination unit 139 determines that the code amount ratio is larger than the threshold value T, the evaluation threshold value set by the intra controller 133 is adjusted to a large degree to reduce the calculation accuracy of the intra prediction encoding, and the motion The search range is set larger for the search range setting unit 241 for setting the search range for obtaining the motion vector by the detection unit 24, and the search accuracy is set to 1/2 for the search accuracy setting unit 242 for setting the search accuracy. A determination signal for obtaining a highly accurate motion vector such as a pixel or a 1/4 pixel is output.
Here, the value of the threshold T is set to 0.6, for example. That is, in relation to the code amount control, the threshold value T is increased when it is desired to allocate more code amount to the P picture than the I picture, and the threshold value T is decreased in the opposite case. In normal encoding, the ratio of the code amount of the I picture and P picture is, for example, about 2: 1. If you want to encode with high coding efficiency and high coding efficiency for an input image with many high-frequency components and uniform motion, set the threshold T to a small value. When it is desired to increase the encoding efficiency for a simple image, the threshold value T is set large. There are a method of setting the threshold T by an operation button and a method of adaptively setting by detecting the property of the input image.

With reference to FIG. 4, the process flow of the image coding apparatus 1 will be described.
First, in S (step) 51, for example, an input image of 1 GOP is encoded. The code amount of I, P, and B pictures obtained by encoding in S52 is acquired. In S53, the code amount ratio is obtained from the code amount Ri of the I picture related to the intra prediction encoding and the average code amount Rp of the P and B pictures related to the inter prediction encoding. The value of T ₄ is the standard value evaluation threshold for intra-coded in S54, if the code amount ratio is detected as the threshold T is smaller than changes in T _4S than T ₄ is a small value To set. In S55, the search range on the reference image for obtaining the motion vector is reduced, and the accuracy of the motion vector is set to a large value.
Obtained in S53, the value of T ₄ evaluation threshold for intra-coding is a standard value in S56 if the code amount ratio is detected as not smaller than the threshold value T, a value greater than the standard value Change and set to T _4L . In S57, the search range on the reference image for obtaining the motion vector is set to a range larger than the standard, and the accuracy of the motion vector is changed and set to a standard value or a value smaller than the standard. The processing from S51 is repeated until it is detected in S58 that there is no input image.

With reference to FIG. 5, the macro block intra prediction will be described.
The intra prediction of a macro block of 16 × 16 pixels shown as mode 0 in the figure is coded by predicting in the vertical direction with reference to 16 pixels adjacent to the upper side of the macro block and indicated by halftone dots. This is an encoding mode when performing encoding. Mode 1 is an encoding mode when encoding is performed by predicting in the horizontal direction with reference to 16 pixels adjacent to the left side of the macroblock. Mode 2 is a coding mode in the case where prediction is performed with reference to the luminance level average values of 32 pixels adjacent to the upper and left sides of the macroblock. Mode 3 is an encoding mode when encoding is performed by predicting in the upper left diagonal direction with reference to 32 pixels adjacent to the upper and left sides of the macroblock.

For the encoding of the macroblock, the prediction value of the macroblock is obtained for each of the cases encoded in the above four prediction modes, and the difference value between the next obtained prediction value and the original image is obtained as a prediction error. Each prediction evaluation value is obtained from the prediction error value. A predicted evaluation value of the mode giving the smallest value of the four predictive evaluation value obtained is compared with the macroblock coding threshold T ₁₆ that is determined in advance, smaller than the macroblock coding threshold T ₁₆ If there is a prediction mode that gives a prediction evaluation value, the macroblock is encoded according to the prediction mode. If the mode which gives the small prediction evaluation value than the macroblock coding threshold T ₁₆ is not present, it performs intra prediction coding on sixteen block images obtained by dividing the macroblock for each 4 × 4 pixels.

Intra prediction coding of a block image will be described with reference to FIG.
In mode 0 shown in the figure, encoding is performed by predicting in the vertical direction from the pixels A to D adjacent to the upper side of the 4 × 4 pixel block. In mode 1, encoding is performed by predicting in the horizontal direction from the pixels I to L adjacent to the left side of the block. In mode 2, predictive encoding is performed using the average value of the luminance levels of the pixels A to D and the pixels I to L. In mode 3, prediction is performed in the lower left direction using the pixels B to D adjacent to the upper side of the block and the pixels E to H on the right side thereof. In mode 4, prediction is performed in the lower right direction from the pixels A to C, M, and I to K. In mode 5, prediction is performed in the vertical direction further from the lower right direction indicated by the arrow from the pixels A to D, M, and I to K. Similarly, prediction directions of modes 6, 7, and 8 are set. There are nine prediction modes in total for the prediction mode of a 4 × 4 pixel block.
There are a total of 16 macroblocks, 4 in the vertical direction and 4 in the horizontal direction, and the intra prediction processing in modes 0 to 8 is performed for each of these blocks. Prediction evaluation values in a mode in which the smallest prediction evaluation value can be obtained by these prediction processes are aggregated for each block, and a prediction evaluation value in the case of intra-coding with a block of 4 × 4 pixels is obtained.

With reference to FIG. 7, the intra prediction encoding process will be described.
First, in S61, a prediction evaluation value is obtained for the macroblock composed of 16 × 16 pixels in the four prediction modes shown in FIG. 5, and the prediction evaluation value which is the smallest value among the four prediction evaluation values is minimized. Obtained as an evaluation value. In S62, the minimum predicted evaluation value or less or not than the macroblock coding threshold T ₁₆ is detected. If there is a prediction mode smaller than the threshold T _{16, the} macroblock intra prediction encoding is performed in S64 in a mode that gives the prediction value. If the small prediction mode is detected as absent than the threshold value T ₁₆ in S62, performs intra prediction processing by the prediction mode of the block of 4 × 4 pixels shown in FIG. 6 S70. In S63, the total prediction evaluation value obtained by summing the prediction evaluation values of the 16 blocks is compared with the minimum prediction evaluation value obtained in S61. When the minimum prediction evaluation value is smaller, the mode that gives the prediction value in S64 is selected as the mode used for intra prediction encoding. When the minimum prediction evaluation value is larger, in S65, intra prediction encoding is performed for each block of 4 × 4 pixels, and a prediction mode that gives the minimum prediction evaluation value for each of the 16 blocks is selected. .

The 4 × 4 pixel (block) intra prediction process will be described with reference to FIGS. In addition, the same code | symbol is attached | subjected about the same process.
The flow of 4 × 4 pixel intra prediction processing (A) illustrated in FIG. 8A is an example in which prediction processing is performed using only three modes 0 to 2 among nine prediction modes. . That is, a predicted evaluation value is obtained for modes 0 to 2 in S71. In S72, the mode that gives the minimum predicted evaluation value among the predicted evaluation values is selected as the prediction mode.
The flow of 4 × 4 pixel intra prediction processing (B) illustrated in FIG. 8B is an example in which prediction processing is performed using only five modes 0 to 4 among nine prediction modes. . In S73, prediction evaluation values are obtained for modes 0 to 4. In S72, the mode that gives the minimum predicted evaluation value among the predicted evaluation values is selected as the prediction mode.

The flow of 4 × 4 pixel intra prediction processing (C) shown in FIG. 9 is an example in which prediction processing is performed using all nine prediction modes. That is, a predicted evaluation value is obtained for modes 0 to 2 in S71. In S74, the prediction evaluation value which gives the smallest value among the predicted evaluation value obtained by intra prediction detects or smaller than the evaluation threshold value T ₄ is a standard threshold using only three modes. If it is detected that the value is smaller than the evaluation threshold value T ₄ , a mode that gives the smallest prediction evaluation value among the prediction evaluation values is selected as the prediction mode in S 72. If it is detected in S74 that it is not small, predicted evaluation values are obtained for modes 3 and 4 in S75. In S74, the predictive evaluation value which gives the smaller value among the predicted evaluation value obtained for detecting or smaller than the evaluation threshold value T _4. When detected as a smaller evaluation threshold T ₄ proceeds to S72, selects the mode giving the smallest prediction evaluation value among the predicted evaluation value obtained by the mode 0-4. If it is detected in S74 that it is not small, a predicted evaluation value is obtained for modes 5 to 8 in S76. Thereafter, in S72, a mode that gives the smallest predicted evaluation value among the predicted evaluation values obtained in modes 0 to 8 is selected.
Here, an example in which the prediction modes of modes 0 to 8 are all prediction modes has been described. That is, the prediction modes of modes 0 to 2 shown in S71 are the first candidate mode group, and the prediction modes of modes 3 and 4 shown in S75 are added to the first candidate mode group, so that the second candidate mode group becomes the first candidate mode group. It is formed. The first candidate mode group includes one or more prediction modes, and one or more prediction modes are added to the first candidate mode group to form a second candidate mode group. The number of all modes may be an arbitrary number of 3 or more, and the intra prediction operation can be performed in the same manner as described above.

The 4 × 4 pixel intra prediction process (D) illustrated in FIG. 10 is an example in which the prediction process is performed using all nine prediction modes and detecting whether or not to perform field coding.
In S71, a prediction evaluation value is obtained for modes 0 to 2. In S74, the predicted evaluation value of the smallest value among the predicted evaluation value obtained detects or smaller than the evaluation threshold value T _4. If it is detected that the value is smaller than the evaluation threshold T _{4, the} mode that gives the minimum prediction evaluation value is selected as the prediction mode in S 72. If it is detected in S74 that it is not small, it is detected in S77 whether or not to perform field coding. If it is detected that field coding is to be performed, predicted evaluation values are obtained for modes 6 and 8 in S78. In S74, the predicted evaluation value of the smallest value among the predicted evaluation value obtained detects or smaller than the evaluation threshold value T _4. If it is detected as small, the process proceeds to S72, and if it is detected as not small, a predicted evaluation value is obtained for modes 3 and 4 in S75. In S74, the predicted evaluation value of the smallest value among the predicted evaluation value obtained detects or smaller than the evaluation threshold value T _4. If it is smaller, the process proceeds to S72. If it is not smaller, a predicted evaluation value is obtained for modes 5 and 7 in S79.

If it is detected in S77 that field coding is not performed, prediction evaluation values are obtained for modes 3 and 4 in S75. In S74, the predictive evaluation value which gives the smaller value among the predicted evaluation value obtained for detecting or smaller than the evaluation threshold value T _4. If it is detected that it is smaller than the evaluation threshold T _{4, the} process proceeds to S 72. If it is detected that it is not smaller, a predicted evaluation value is obtained for modes 5 to 8 in S 76.
In S72, a mode that gives the smallest predicted evaluation value among all the modes for which the predicted evaluation value is obtained is selected. Intra coding is performed in a mode in which the smallest prediction evaluation value is obtained. Compared to encoding for all modes, and intra encoding using the encoding result of the mode that gives the smallest code amount, encoding performed while obtaining a prediction evaluation value, The speed of the encoding operation can be increased, for example, it can be halved. Compared to intra coding performed while obtaining prediction evaluation values for all modes, a coded output having a comparable code amount can be obtained.

Here, the value of the evaluation threshold T ₄ is changed according to the definition of the input image, while the data amount of the encoded output obtained by the intra encoding as described in FIG. It is also changed by the magnitude relationship of the code amount ratio with the data amount of the encoded output obtained by inter encoding. When it is necessary to perform intra coding with high coding efficiency, T _4S is set as an evaluation threshold smaller than the value of the evaluation threshold T ₄ , and when it is necessary to reduce the calculation speed more than coding efficiency, the evaluation threshold T By setting as T _4L , which is an evaluation threshold value larger than the value of ₄ , intra coding can be executed while selecting coding efficiency and the amount of calculation at the time of coding. When the image encoding device is portable and battery-operated, the remaining battery level is detected. When the remaining battery level is low, the evaluation threshold value T ₄ is set to T _4L and further to T Change to T _4LL , which is larger than _4L . Similarly, the value of the evaluation threshold T ₄ is changed while detecting the remaining amount of the recording medium. When the remaining amount of the medium is reduced, the value of the evaluation threshold T ₄ is changed to T _4S and further to T _4SS having a value smaller than T _4S .

  As described above, according to the image encoding device shown in the present embodiment, one prediction mode is selected from a plurality of prediction modes prepared in advance, and a block image is intra prediction encoded in the selected prediction mode. Encoding means (112) for performing prediction error detection means for detecting a difference between a prediction image obtained when the block image is intra-predicted in an arbitrary prediction mode and a block image as a prediction error corresponding to the arbitrary prediction mode (113) and an operation setting means that detects the number of pixels of the input image and sets an operation method to perform a standard operation when the number of pixels is within a predetermined number and to perform a high-definition operation when the number of pixels exceeds the predetermined number ( 132) and one or more prediction modes selected in advance among the plurality of prediction modes as a first candidate mode group, and one or more other predictions in the plurality of prediction modes in the first candidate mode group A mode group setting means (111) for adding a mode and setting as a second candidate mode group, and a first candidate for setting a block image by the mode group setting means when set as a standard action by the action setting means The prediction error in each prediction mode of the mode group is detected by the prediction error detection means, and when there is a prediction error below a predetermined threshold in each detected prediction error, one prediction mode that gives the prediction error is set to one. When there is no prediction mode that gives a prediction error below a predetermined threshold for each detected prediction error while selecting as a prediction mode, detection of prediction errors corresponding to all of the plurality of prepared prediction modes is detected. A prediction mode that detects the smallest prediction error among the detected prediction errors is selected as one prediction mode, When the high-definition operation is set by the prediction error detection means, the prediction error in each of the prediction modes of the first candidate mode group set by the mode group setting means is detected by the block image, and each detected When the prediction error includes a prediction error that is larger than a predetermined threshold and is not larger than the enlargement threshold, the prediction mode that gives the prediction error is selected as one prediction mode, while each detected prediction error is smaller than or equal to the enlargement threshold. Is not detected, the prediction error in each of the prediction modes of the second candidate mode group set by the mode group setting unit is detected by the prediction error detection unit, and the smallest of the detected prediction errors is detected. The encoding means is controlled so that a prediction mode giving a prediction error is selected as one prediction mode and intra prediction encoding is performed. Coding control means (133) that can effectively reduce the number of prediction modes, reduce the amount of calculation during coding, reduce power consumption, and enable coding with high coding efficiency. An image encoding device can be realized.

It is a block diagram which shows the structural example of the image coding apparatus which concerns on implementation of this invention. It is a figure which shows the structural example (the 1) of the principal part of the image coding apparatus which concerns on implementation of this invention. It is a figure which shows the structural example (the 2) of the principal part of the image coding apparatus which concerns on implementation of this invention. It is the figure which showed the operation | movement of the image coding apparatus based on implementation of this invention with the flowchart. It is a figure for demonstrating each mode of the intra prediction of a 16x16 pixel block. It is a figure for demonstrating each mode of 4 * 4 pixel block intra prediction. It is the figure (the 1) which showed the intra prediction operation | movement which concerns on implementation of this invention with the flowchart. It is the figure (the 2) which showed the intra prediction operation | movement which concerns on implementation of this invention with the flowchart. It is the figure (the 3) which showed the intra prediction operation | movement which concerns on implementation of this invention with the flowchart. It is the figure (the 4) which showed the intra prediction operation | movement which concerns on implementation of this invention with the flowchart.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image coding apparatus 11 Intra prediction part 12 Subtractor 13 Encoding control part 13a Intra control part 13b Inter control part 14 Conversion part 15 Quantization part 16 Intropy encoding part 17 Inverse quantization part 18 Inverse conversion part 19 Adder 21 Loop filter 22 Frame memory 23 Motion compensation unit 24 Motion detection unit 25 Switch 111 Prediction mode setting unit 112 Intra prediction execution unit 113 Prediction error calculator 114 Encoded data acquisition unit 131 Pixel number detector 132 Pixel number determination unit 133 Intra controller 135 Code amount acquisition unit 136 Inter code amount storage unit 137 Intra code amount storage unit 138 Code amount comparison unit 139 Condition determination unit 241 Search range setting unit 242 Search accuracy setting unit 243 Motion detection execution unit

Claims (2)

In an image encoding device that divides an input image into a predetermined number of vertical and horizontal pixels and encodes block images obtained by the division while sequentially performing intra prediction,
A mode group in which two or more prediction modes selected in advance among a plurality of prediction modes are set as a first candidate mode group, and another prediction mode is added to the first candidate mode group and set as a second candidate mode group Setting means;
An operation setting means for detecting the number of pixels of the input image and setting an operation to be standard when the number of pixels is within a predetermined number, and to perform high-definition when exceeding the predetermined number;
A prediction image obtained by intra-predicting the block image in each prediction mode of the first candidate mode group set by the mode group setting unit and the block image when set as the standard by the operation setting unit The prediction error that is the difference between the prediction error and the prediction error corresponding to each prediction mode is obtained, and when there is a prediction error that is equal to or less than a predetermined threshold in each of the obtained prediction errors, one prediction mode that gives the prediction error is provided. On the other hand, if there is no prediction error equal to or less than the predetermined threshold in each of the obtained prediction errors, the respective prediction errors corresponding to all of the plurality of prediction modes are obtained, and each of the obtained prediction errors is calculated. Select one of the prediction modes that gives the smallest prediction error,
The prediction image and the block obtained by intra-predicting the block image in each prediction mode of the first candidate mode group set by the mode group setting unit when the operation setting unit sets the high definition A prediction error that is a difference from the image is obtained as a prediction error corresponding to each prediction mode, and when each of the obtained prediction errors includes a prediction error equal to or smaller than the enlargement threshold that is larger than the predetermined threshold, When one prediction mode that gives a prediction error is selected and there is no prediction error equal to or less than the expansion threshold in each obtained prediction error, each prediction error corresponding to the prediction mode of the second candidate mode group And mode selection means for selecting one prediction mode that gives the smallest prediction error among the obtained prediction errors,
Encoding means for obtaining an encoded signal by intra prediction encoding the block image using one prediction mode selected by the mode selection means;
An image encoding apparatus comprising: Image coding that divides an input image into a predetermined number of pixels vertically and horizontally and encodes the block image obtained by dividing the image while sequentially predicting it according to a predetermined rule while performing intra prediction and motion prediction. In the device
A mode group setting means for setting two or more preselected prediction modes as a candidate mode group among a plurality of prediction modes;
A search range for obtaining a motion vector by searching for a matching position of the block image on the reference image obtained by encoding and decoding an input image as a reference image, and a standard range and the standard range Search range setting means for setting as a larger expansion range,
The ratio of the code amount of the encoded signal generated by motion prediction immediately before the input image to the code amount of the encoded signal generated by intra prediction immediately before the input image is a code amount ratio. Code amount ratio detecting means for detecting
When the code amount ratio detected by the code amount ratio detection unit is larger than a predetermined threshold, the block image is intra predicted in each prediction mode of the candidate mode group set by the mode group setting unit, A difference between each prediction image obtained by intra prediction and the block image is obtained as a prediction error corresponding to each prediction mode, and a prediction mode that gives the smallest prediction error among the obtained prediction errors is determined as the intra prediction mode. While selecting as a prediction mode when predicting, while selecting the expansion range set by the search range setting means as a motion vector search range to be obtained when the motion prediction,
When the code amount ratio detected by the code amount ratio detection unit is equal to or less than the predetermined threshold, the block image is intra-predicted in all of the plurality of prediction modes, and each prediction obtained by intra prediction A difference between an image and the block image is obtained as a prediction error corresponding to each prediction mode, and a prediction mode that gives the smallest prediction error among the obtained prediction errors is selected as a prediction mode for the intra prediction. And a selection means for selecting the standard range selected by the search range setting means as a search range for a motion vector obtained when the motion prediction is performed,
Intra coding means for obtaining an intra prediction coded signal by intra prediction coding the block image using the prediction mode selected by the selection means;
Motion prediction encoding means for obtaining a motion prediction encoded signal by performing motion prediction encoding using a motion vector determined by the search range selected by the selection means;
Entropy encoding means for entropy encoding the intra prediction encoded signal obtained by the intra encoding means and the motion prediction encoded signal obtained by the motion prediction encoding means to generate the encoded signal;
An image encoding apparatus comprising:

Images