JP4359184B2 - Prediction information / quantized value control compression encoding apparatus, prediction information / quantization value control compression encoding program - Google Patents

Description

  The present invention relates to a prediction information / quantization value control compression encoding apparatus, a prediction information / quantization value control compression encoding program, and a prediction information / quantization value control compression encoding method for compressing and encoding a moving image.

  In general, it is known that when a moving image is compression-encoded, flickering (degradation of image quality due to luminance or color variation) may occur in a decoded image obtained by decoding the encoded image that has been compression-encoded. .

  Conventionally, for example, a “flicker reduction circuit” (see Patent Document 1) has been disclosed as a means for reducing (reducing) the flicker (preventing flicker interference). This flicker reduction circuit reduces the flicker at the time of quantization in A / D conversion, and the current digital signal after A / D conversion of the input video signal is latched by the latch circuit and recorded in the field memory. Then, the recorded current digital signal and the previous digital signal after A / D conversion of the previously input video signal are subtracted to obtain an absolute value, and output after determining whether the absolute value is smaller than 1 By doing so, flicker is reduced.

  Here, a quantized signal (signal waveform) obtained by quantizing an input signal (digital signal) will be described with reference to FIGS. FIG. 9 shows a graph before a signal waveform is drawn. In FIG. 9, the horizontal axis is a time axis, and the horizontal axis is a digital signal delimiter (the first sample from the vertical axis is the first sample, the next sample is the second sample, ...)). Also, the vertical axis indicates the value that can be taken by the digital signal, the dotted line in FIG. 9 shows the value that can be taken by the quantized signal, and the alternate long and short dash line in FIG. 9 shows the value after quantization (0 or 1) shows the changing boundary value.

  FIG. 10A shows an example of an input signal (digital signal), and FIG. 10B shows the result of simple quantization of the input signal shown in FIG. 10A (after quantization). Signal). As shown in FIG. 10B, the quantized signal has a larger fluctuation (amplitude difference) than the input signal, and flicker interference occurs.

  When the signal waveform of FIG. 10B is input to a conventional “flicker reduction circuit”, the signal waveform shown in FIG. 10C is obtained. In FIG. 10C, the second sample outputs the same value as the first sample because the difference from the first sample is 1. As a result, the conventional “flicker reduction circuit” can simply prevent flicker interference that occurs when quantization is performed (FIG. 10B).

  By the way, when compressing and encoding a moving image, if a compression encoding method that performs quantization in the process of processing the residual signal obtained by subtracting the prediction signal from the video signal (original signal) of the original image is compressed. Variations in luminance and color (flicker) appear in the decoded image obtained by decoding the encoded image, and image degradation of the decoded image occurs. This deterioration in image quality occurs periodically (in the period of an intra frame), and particularly appears when the compression rate of the encoded image is high.

JP-A-8-163470 (paragraphs 0004 to 0008, FIG. 1)

  However, the compression coding method using the prediction signal is not supported by the conventional flicker reduction circuit, and even if the flicker reduction circuit is applied to the compression coding method using the prediction signal, flicker interference can be prevented. There is a problem that you can not.

  Here, with reference to FIG. 11, a quantized signal (signal waveform) when a conventional flicker reduction circuit is applied to a compression encoding method using a prediction signal will be described. The graph shown in FIG. 11 has the same view as the graph shown in FIG. In FIG. 11, in order to simplify the description, conversion processing performed in general video compression is omitted. That is, description will be made assuming that the residual signal obtained by subtracting the prediction signal from the video signal (original signal) of the original image is quantized. Incidentally, at the time of decoding, the quantized quantized value is inversely quantized, and the prediction signal is added to obtain a video signal (original signal) of the original original image.

FIG. 11A shows an original signal, a prediction signal, and a residual signal obtained by subtracting the prediction signal from the original signal.
FIG. 11B shows a quantized signal obtained by quantizing the residual signal of FIG. 11A as an inversely quantized signal obtained by further inverse quantization. When the conventional flicker reduction circuit is used, the difference between the quantized value (current digital signal) and the output of the previous frame (previous digital signal after A / D conversion of the previously input video signal) is 1 (absolute When the value is 1), the current quantized value is used as the output of the previous frame.

  In FIG. 11B, when focusing on the second sample (convex portion), the difference from the first sample is 2, so the second sample is output as it is (that is, the first sample is copied). Not) Since the difference between the third sample and the second sample is 2, the third sample is output as it is (that is, the second sample is not copied). That is, in the example shown in FIG. 11B, even if a conventional flicker reduction circuit is applied, the result is not different from the case of simple quantization.

  FIG. 11C shows a decoded signal obtained by dequantizing the output of FIG. 11B and adding the prediction signal. Although the original signal (see FIG. 11 (a)) is the same as the input signal (video signal) shown in FIG. 10 (a), the value of the decoded signal (the signal waveform of the decoded signal) varies greatly. Flicker interference has occurred. That is, as described above, the compression encoding method using the prediction signal (quantizing the residual signal obtained by subtracting the prediction signal from the original signal) has the effect of preventing flicker interference in the conventional flicker reduction circuit. Does not occur.

  Further, another compression coding method (for example, H.264 method) using a prediction signal is disclosed in Japanese Patent Application No. 2003-197008 (flicker reduction quantizer, flicker reduction quantization method, and program thereof). In addition to controlling the quantization value, quantization / inverse quantization processing is performed on the prediction signal. However, the quantization / inverse quantization processing of the prediction signal is performed in H.264. A decoder compliant with the H.264 standard has a problem that it cannot be decoded. That is, the quantization / inverse quantization processing of the prediction signal is performed by H.264. It cannot be executed because the process deviates from the H.264 system.

  Therefore, in the present invention, a prediction information / quantization value control compression encoding apparatus, prediction information / quantization value control compression, which can solve the above-described problems and prevent flicker interference in a compression encoding method using a prediction signal. It is an object of the present invention to provide an encoding program and a prediction information / quantized value control compression encoding method.

  In order to solve the above-described problem, the prediction information / quantization value control compression encoding apparatus according to claim 1 is configured to extract an image of a predetermined block included in a moving image in a motion of the predetermined block or a neighborhood of the predetermined block. Controlling the combination of multiple prediction information generated based on prediction information predicted from the motion of a region and a quantization value obtained by quantizing the moving image and a corresponding quantization value corresponding to the multiple prediction information; A prediction information / quantization value control compression encoding apparatus for compressing and encoding an image, comprising: prediction information estimation means, combination generation means, prediction signal generation means, quantization value output means, and local decoded image output means And compressed data output means, selection means, and output means.

According to this configuration, the prediction information / quantization value control compression encoding apparatus estimates the prediction information based on the reference image and the block image by the prediction information estimation unit, and outputs the prediction information to the combination generation unit. by, for generating a plurality of multiple prediction information generated based on the prediction information, and corresponding quantized values corresponding to the plurality prediction information. The prediction information defines a mode (operation method) and a value for generating a prediction signal. In the H.264 system, it is defined that a prediction signal is generated using a motion vector value in a mode called “motion compensation”, or a prediction signal is used using a value indicating a prediction direction in a mode called “intra prediction”. It is stipulated to generate.

  Subsequently, the prediction information / quantization value control compression encoding apparatus generates a block of a predetermined region in the reference image as a prediction signal based on the multiple prediction information by the prediction signal generation unit, and the quantization value output unit A quantized value obtained by encoding a subtraction signal obtained by subtracting the prediction signal from the image is output to the combination generating means. The prediction signal is created as a result of prediction, and is, for example, a luminance signal. A block image is a block in a predetermined area of a frame to be compression-encoded. The local decoded image is obtained by decoding a once-encoded block image.

Then, the prediction information and quantization value control compression encoding apparatus, the compressed data output means, to compress the combination of a plurality prediction information generated by the combination generating means and a corresponding quantized value outputted as compressed data, the local decoded The quantized value in the compressed data is decoded by the image output means and output as a local decoded image .

  Then, the prediction information / quantization value control compression encoding apparatus uses the selection unit to determine the luminance difference between the block image immediately before decoding and the local decoded image obtained by decoding the block image immediately before the predetermined region in the frame immediately before the block image. Compressed data that becomes less than a preset value is selected. Note that the set value is a value at which the luminance difference between the decoded block image immediately before decoding and the local decoded image becomes sufficiently small, that is, a value at which the luminance variation cannot be visually recognized (recognized) as image quality degradation.

  Then, the prediction information / quantized value control compression encoding apparatus holds the compressed data output by the compressed data output means by the output means, and outputs the compressed data selected by the selection means as the optimum compressed data. In other words, the output means has a buffering function, and outputs the compressed data selected by the selection means as optimum compressed data.

The prediction information / quantization value control compression encoding apparatus according to claim 2 predicts an image of a predetermined block included in a moving image based on a motion of the predetermined block or a motion of a region near the predetermined block in advance. Prediction that compresses and encodes the moving image by controlling the combination of the limited number of prediction information and the quantized value obtained by quantizing the moving image and the corresponding quantized value corresponding to the multiple prediction information An information / quantized value control compression encoding apparatus, wherein a combination generating means, a prediction signal generating means, a quantized value output means, a local decoded image output means, a compressed data output means, a selection means, and an output Means.

  According to this configuration, the prediction information / quantized value control compression encoding apparatus generates multiple prediction information and a corresponding quantized value corresponding to the multiple prediction information by the combination generation unit. Subsequently, the prediction information / quantization value control compression encoding device generates a block of a predetermined region in the reference image as a prediction signal based on the multiple prediction information by the prediction signal generation unit, and the quantization value output unit A quantized value obtained by encoding a subtraction signal obtained by subtracting the prediction signal from the block image is output to the combination generating means.

Then, the prediction information and quantization value control compression encoding apparatus, the compressed data output means, to compress the combination of a plurality prediction information generated by the combination generating means and a corresponding quantized value outputted as compressed data, the local decoded The quantized value in the compressed data is decoded by the image output means and output as a local decoded image .

  Then, the prediction information / quantization value control compression encoding apparatus uses the selection unit to determine the luminance difference between the block image immediately before decoding and the local decoded image obtained by decoding the block image immediately before the predetermined region in the frame immediately before the block image. Compressed data that becomes less than a preset value is selected. Then, the prediction information / quantized value control compression encoding apparatus holds the compressed data output by the compressed data output means by the output means, and outputs the compressed data selected by the selection means as the optimum compressed data.

That is, this prediction information / quantization value control compression encoding apparatus is a mode in which prediction information is not estimated in the prediction information / quantization value control compression encoding apparatus according to claim 1. That is, when the number of pieces of multiple prediction information generated based on the prediction information is limited, the combination generation unit generates a plurality of pieces of multiple prediction information without requiring input of prediction information. In this case, it is necessary to input in advance to the combination generation means that a plurality of pieces of prediction information are limited (a mode or a value for generating a prediction signal is small). For example, H.M. In the H.264 intra prediction, since the standard defines that there are only several types of prediction information, this may be set (input) in advance in the combination generating means.

  The prediction information / quantization value control compression encoding apparatus according to claim 3 is the prediction information / quantization value control compression encoding apparatus according to claim 1 or 2, wherein the selection unit is configured to determine whether the luminance difference is The compressed data that is minimized is selected.

  According to such a configuration, the prediction information / quantized value control compression encoding apparatus selects the compressed data that minimizes the luminance difference by the selection unit, thereby decoding the compressed data (optimum compressed data). Flicker interference can be reduced most.

4. prediction information and quantization value control compression encoding apparatus described in the prediction information and quantized value control compression encoding apparatus according to claim 1, wherein the combination generating means, the block image, the intra prediction When the prediction information output by the prediction information estimation unit is input, only the prediction information is output as the multiple prediction information, and the quantization value output unit outputs the prediction information. When a quantized value is input, only the quantized value is output as the corresponding quantized value .

According to such a configuration, when the prediction information / quantization value control compression encoding apparatus can select and process either intra prediction or inter prediction for a block image in a frame , for example, by a combination generation unit. the intra prediction by processing only the rows naive lock image, when decoding the optimal compression data outputted from the output means, it is possible to reduce the flicker disturbance. Incidentally, intra prediction is a prediction mode in which a prediction signal is created from the same frame as the frame to be encoded, and inter prediction is a prediction in which a prediction signal is created from a frame different from the frame to be encoded. It is a mode.

6. The prediction information / quantization value control compression encoding apparatus according to claim 5 is the prediction information / quantization value control compression encoding apparatus according to claim 1 , wherein the combination generation unit is configured to transmit the block image as an intra frame. If the prediction information output by the prediction information estimation means is input, the prediction information output means outputs only the prediction information as the multiple prediction information, and the quantization value output means outputs the prediction information. When a quantized value is input, only the quantized value is output as the corresponding quantized value .

According to such a configuration, the prediction information / quantization value control compression encoding apparatus decodes the optimum compressed data output from the output unit by processing only the block image in the intra frame by the combination generation unit. In addition, flicker interference can be reduced. An intra frame is a frame in which inter prediction is prohibited in all blocks in the intra frame.

The prediction information / quantization value control compression encoding apparatus according to claim 6 is the prediction information / quantization value control compression encoding apparatus according to claim 1 , wherein brightness of the block image and the block image immediately before decoding is determined. Block comparison means for comparing the difference and outputting a comparison result, wherein the combination generation means is configured to output the prediction information when a luminance difference included in the comparison result output from the block comparison means is greater than a predetermined value. When the prediction information output by the estimation means is input, only the prediction information is output as the plurality of prediction information, and when the quantization value output by the quantization value output means is input, the corresponding Only the quantized value is output as a quantized value .

According to such a configuration, the prediction information / quantization value control compression encoding apparatus compares the luminance difference between the block image and the block image immediately before decoding by the block comparison unit, and the block generation unit immediately before decoding by the combination generation unit. By processing only the block image whose luminance difference from the block image is a certain value or less, flicker interference can be reduced when the optimal compressed data output from the output means is decoded. That is, when the luminance difference between the block image and the block image immediately before decoding is larger than a certain value, it is assumed that there is a luminance variation in the original moving image (original image). Thus, it is not necessary to perform processing for reducing flicker interference, and the processing speed can be improved.

  The prediction information / quantization value control compression encoding apparatus according to claim 7 is the prediction information / quantization value control compression encoding apparatus according to any one of claims 1 to 6, wherein the selection unit includes: Instead of the block image immediately before decoding, the luminance difference between the intra block image immediately before decoding and the local decoded image obtained by decoding the immediately preceding intra block image in the same region as the predetermined region in the intra frame immediately before the block image is minimized. The compressed data to be selected is selected.

  According to such a configuration, the prediction information / quantization value control compression encoding apparatus, when the bit rate is low, is not the block image of the same area as the predetermined area of the immediately preceding frame, but immediately before decoding. Optimal compression output from the output means by selecting the compressed data that minimizes the luminance difference between the intra-block image immediately before decoding and the local decoded image obtained by decoding the previous intra-block image in the same area as the predetermined area in the intra frame Flicker interference can be reduced when data is decoded.

  The prediction information / quantization value control compression encoding apparatus according to claim 8 is the prediction information / quantization value control compression encoding apparatus according to any one of claims 1 to 7, wherein the selection unit includes: The compressed data is selected based on a difference in hue in addition to the luminance difference.

  According to such a configuration, the prediction information / quantized value control compression encoding apparatus outputs the output from the output unit by selecting the compressed data based on the difference in hue as well as the difference in luminance value by the selection unit. When the optimum compressed data is decoded, flicker interference can be reduced.

  The prediction information / quantization value control compression encoding program according to claim 9 predicts an image of a predetermined block included in a moving image from a motion of the predetermined block or a motion of a region near the predetermined block. Controlling a combination of multiple prediction information generated based on information and a quantization value obtained by quantizing the moving image and a corresponding quantization value corresponding to the multiple prediction information, and compressing and encoding the moving image In addition, the computer is configured to function as prediction information estimation means, combination generation means, prediction signal generation means, quantized value output means, local decoded image output means, compressed data output means, selection means, and output means.

According to this configuration, the prediction information / quantization value control compression encoding program estimates the prediction information based on the reference image and the block image by the prediction information estimation unit, and outputs the prediction information to the combination generation unit. To generate a plurality of pieces of prediction information generated based on the prediction information and a corresponding quantized value corresponding to the pieces of prediction information. Subsequently, the prediction information / quantized value control compression encoding program generates a block of a predetermined region in the reference image as a prediction signal based on the plurality of pieces of prediction information by the prediction signal generation unit. The prediction information / quantized value control compression encoding apparatus outputs a quantized value obtained by encoding a subtraction signal obtained by subtracting the prediction signal from the block image to the combination generating unit by the quantized value output unit, and outputs compressed data. Means for compressing the combination of the multiple prediction information generated by the combination generating means and the corresponding quantized value and outputting the compressed data, and decoding the quantized value in the compressed data by the local decoded image output means Output as a local decoded image .

  Then, the prediction information / quantization value control compression encoding program uses the selection unit to calculate the luminance difference between the block image immediately before decoding and the local decoded image obtained by decoding the block image immediately before the predetermined region in the frame immediately before the block image. Is selected to be less than a preset set value, and the output means outputs the compressed data selected by the selection means from among the compressed data output by the compressed data output means as the optimum compressed data.

According to claim 1, 2, and 9 SL placing the invention, flicker disturbance caused slight brightness difference between the immediately preceding block image in the frame immediately preceding the block image is enlarged by the quantization processing (image quality deterioration), It can be reduced by selecting a combination of the multiple prediction information and the corresponding quantized value for which the luminance difference is less than the setting. In other words, flicker interference can be prevented in the compression encoding method using the prediction signal.

  According to the third aspect of the present invention, flicker interference (deterioration in image quality) caused by a slight luminance difference between the block image and the immediately preceding block image being enlarged by the quantization process is detected with the prediction information and the quantum that minimize the luminance difference. This can be reduced by selecting a combination with the digitization value.

According to the fourth aspect of the present invention, the block image in the frame to perform by selecting one of the intra prediction and inter prediction, the inter prediction line naive locked image, the variation in brightness is less likely to occur, intra predict by processing only the rows naive locked image, it is possible to sufficiently suppress the variation in luminance, it is possible to prevent flicker disturbance. Particularly when the bit rate is low, the row naive lock image inter prediction, since in many cases a copy of the previous frame, without applying the inter prediction in the row naive lock image, row intra prediction naive By applying only the lock image, the effect of suppressing flicker interference can be sufficiently obtained.

  According to the fifth aspect of the present invention, flicker interference can be reduced by processing only the block image in the intra frame without processing the block image in the inter frame. That is, there are few block images which perform intra prediction within an inter frame. In addition, the fact that inter prediction was not performed means that the luminance difference between frames (between the current frame and the previous frame) was large at the stage of the moving image (original image, original signal) (originally large luminance fluctuation). This is not because there was a large luminance fluctuation due to compression encoding, so if only the block image in the intra frame is processed, the effect of suppressing flicker interference can be sufficiently obtained.

  According to the sixth aspect of the present invention, when the luminance difference between the block image to be compression-encoded and the block image immediately before decoding is large, the moving image (original image, original signal) has a large luminance variation. Therefore, if processing other than such a block image is performed, flicker interference caused by compression coding can be reduced.

  According to the seventh aspect of the present invention, when compressing and encoding an intra frame, the prediction information and the prediction information are set so that the luminance difference between the intra frame immediately before decoding and the block image immediately before decoding in the previous intra frame is small. If the combination with the quantized value is selected, the effect of suppressing flicker interference can be sufficiently obtained. That is, when the bit rate is low, when the inter frame is continuous (the inter frame group exists), the luminance value of the block image in the intra frame existing immediately before the inter frame group is retained. This is because there are many (especially when there are many still video portions in a moving image).

  According to the eighth aspect of the present invention, flicker interference can be reduced when the optimum compressed data output from the output means is decoded by selecting the compressed data based on the difference in hue.

Next, embodiments of the present invention will be described in detail with reference to the drawings as appropriate.
Here, three embodiments (from the first embodiment to the third embodiment) of the prediction information / quantized value control compression encoding apparatus will be described. In addition, about operation | movement of the said apparatus, since the structure of each apparatus is based on the structure of 1st embodiment, suppose that only operation | movement of 1st embodiment is demonstrated.

<Configuration of Prediction Information / Quantization Value Control Compression Encoding Device (First Embodiment)>
FIG. 1 is a block diagram of a prediction information / quantization value control compression encoding apparatus (first embodiment). As shown in FIG. 1, the prediction information / quantized value control compression encoding apparatus 1 compresses and encodes a moving image by controlling a combination of prediction information and a quantization value. A video compression means 5, a selection means 7, and an output means 9.

  Although not shown in FIG. 1, the prediction information / quantization value control compression encoding apparatus 1 includes an input processing unit that performs a process of cutting out a block image to be compressed from an input signal that is a moving image, and an output. Variable-length encoding means for further compressing the optimum compressed data output from the means 9, and a frame memory (local decoded image storage means) for holding the local decoded image B ′ (c, p) in the reference image ) Etc.

  The combination generating unit 3 receives the prediction information and the quantized value output from the video compression unit 5 and generates a plurality of prediction information that is a variation of the prediction information, and a corresponding quantized value corresponding to the plurality of prediction information And some of these combinations are output to the video compression means 5.

  Prediction information is for predicting an image signal, and defines the type of prediction method used when generating a prediction signal and the value used for prediction. For example, the types of prediction methods such as motion compensation prediction and pixel value prediction from adjacent pixels, and values such as motion vector values and adjacent pixel values are defined.

  The multiple prediction information is generated based on the prediction information. For example, a specific value (for example, ± 1) set in advance is added to a value used for prediction included in the prediction information. Is. That is, the multiple prediction information can be said to be derived from one prediction information (variation).

  The quantized value is obtained by converting the input signal (original signal) that is a moving image after the conversion process (DCT conversion or the like), and the larger the quantization characteristic value (Q; the greater the amount of information when quantized). The decoded pixel is divided by a value having a property of becoming coarse.

The corresponding quantized value is a quantized value obtained corresponding to multiple prediction information.
A combination refers to a set of multiple prediction information and a corresponding quantized value, and the number of generated combinations differs depending on the type of prediction method included in the prediction information.

  Here, examples of combinations will be described. The corresponding quantization value is selected from three types by adding −1, 0, +1 to the quantization value obtained from the video compression means 5. This improves the balance (performance) between the effect of suppressing flicker interference and the amount of calculation performed by the device 1.

  The multi-prediction information largely depends on the prediction method used in the compression coding method (video compression method) applied in the apparatus 1. For example, when motion compensation prediction is used, values of motion vectors (horizontal direction and vertical direction respectively) ) Plus -1, 0, +1.

  H. For prediction methods that are limited in several ways, such as intra prediction in H.264, all prediction methods may be selected. When the number of combinations is calculated for these two examples (motion compensation prediction and intra prediction in the H.264 system), in the case of motion compensation prediction, 3 (three quantization values) × 3 (horizontal motion) 3 types of vectors) × 3 (3 types of motion vectors in the vertical direction) = 27 types. H. In the case of intra prediction in the H.264 system, 3 (3 quantization values) × 4 (the number of intra 16 × 16 prediction modes in the H.264 system) = 12.

  Note that it is not necessary to output all of these combinations from the combination generating unit 3 to the video compression unit 5, and it is possible to omit and output them appropriately according to the amount of calculation.

  The video compression means 5 selects a block image B (c, p) to be compressed (an image of a block at a predetermined position p in the c-th frame) and a reference image (an image to be referred to when a moving image is compressed). Input and output of prediction information and quantized values and input of plural prediction information and corresponding quantized values between the combination generating means 3 and the compressed data and the local decoded image B ′ (c, p ) (Decoded block image obtained by decoding a block at a predetermined position p in the c-th frame).

  That is, the video compression means 5 compresses the input block image B (c, p) based on the multiple prediction information generated by the combination generation means 3, outputs the encoded quantized value, and generates a combination. Compressed data is generated from the corresponding quantized value output from the means 3 and the plurality of prediction information and output. The video compression means 5 outputs a local decoded image B ′ (c, p) obtained by decoding the compressed data.

  After this, the prediction information / quantization value control compression coding apparatus 1 is a block immediately before decoding, which is a decoded block immediately before existing in the same predetermined area (spatial position) as the block image B (c, p). From the image B ′ (r, p) (part of position p in the r-th frame, in this case, r = c−1 since the r-th frame is immediately before the c-th frame), the local decoded image B ′ (C, p) is subtracted, the combination having the smallest difference (luminance difference) is selected by the selection means 7, and the compressed data generated by this combination (multiple prediction information, corresponding quantized value) is optimally compressed. Output from the output means 9 as data.

  When there are a plurality of combinations that minimize the difference between the local decoded image B ′ (c, p) and the block image B ′ (r, p) immediately before decoding, the one with the better image compression rate and image quality is selected. You just have to choose. Even if the combination with the smallest difference is not selected, a combination with a sufficiently small image difference (less than the set value) may be selected with a better image compression rate or image quality. Incidentally, the compression rate of the image is obtained from the size of the compressed data. For example, the smaller the size of the compressed data, the better the compression rate of the image. The better the image quality, the better the smaller the difference between the local decoded image B ′ (c, p) and the block image B (c, p).

  Here, the video compression means 5 will be described in detail with reference to FIG. 2 which is a block diagram in which the video compression means 5 is expanded (the structure is divided). As shown in FIG. 2, the video compression means 5 includes a prediction information estimation means 11, a prediction signal generation means 13, an encoding means (quantized value output means) 15, and a compressed data generation means (compressed data output means). 17 and a decoding means (local decoded image output means) 19.

The prediction information estimation means 11 estimates prediction information from the block image B (c, p) to be encoded and the reference image. This prediction information is obtained, for example, for the motion vector, the direction of intra prediction, and the like so that the predicted image and the original image are closest. In addition, although illustration was abbreviate | omitted here, this prediction information estimation means 11 includes a motion vector estimation means and an intra prediction estimation means, and the prediction information estimation means 11 calculates | requires prediction information by the existing technique. It is what.

  The prediction signal generation unit 13 generates a prediction signal from the reference image based on the multiple prediction information generated by the combination generation unit 3. The prediction signal is a luminance signal (or color difference signal) generated as a result of prediction based on prediction information. In the compression coding method using prediction information, in order to improve transmission efficiency, a difference between a predetermined area in the current frame and a copy of the predetermined area in the immediately preceding frame is transmitted. Yes.

  By the way, in MPEG-2 and MPEG-4, the prediction signal is copied (copied) from the frame one frame before or after the frame to be encoded (inter prediction).

  The encoding unit 15 encodes a subtraction signal obtained by subtracting the prediction signal generated by the prediction signal generation unit 13 from the input block image B (c, p). The conversion unit 15a and the quantization unit 15b.

The conversion means 15a performs a conversion process on the subtraction signal, and in this embodiment, DCT conversion is adopted. Note that the conversion process includes, for example, Hadamard conversion in addition to DCT conversion.
The quantizing unit 15b quantizes the converted signal converted by the converting unit 15a and outputs the quantized value to the combination generating unit 3 (see FIG. 1) as a quantized value.

  The compressed data generation means 17 is a variable length encoding means (not shown), etc., which is a combination of the prediction information generated by the combination generation means 3 and the corresponding quantized value (information necessary for decoding). It is generated and output as compressed data compressed in step (b).

  The decoding unit 19 decodes the corresponding quantized value included in the compressed data generated by the compressed data generating unit 17, adds the prediction signal, and outputs a local decoded image B ′ (c, p). In this case, an inverse quantization unit 19a and an inverse transformation unit 19b are provided.

The inverse quantization means 19a outputs the corresponding quantized value to the inverse transform means 19b as a transformed signal obtained by inverse quantization.
The inverse transform unit 19b performs inverse transform on the transform signal that is the result of inverse quantization of the corresponding quantized value by the inverse quantizer 19a and outputs it as a decoded signal. The decoding means 19 adds the prediction signal to this decoded signal and outputs it as a local decoded image B ′ (c, p).

  The video compression means 5 does not need to have all the configurations shown in FIG. 2. If the prediction information estimation means 11, the prediction signal generation means 13, and the encoding means 15 are provided, any compression encoding is possible. The method (video compression method) can also be applied.

Returning to FIG. 1, the description of the configuration of the prediction information / quantized value control compression encoding apparatus 1 will be continued.
The selection means 7 decodes the block image B ′ (r immediately before decoding) obtained by decoding the block image B (r, p) of the immediately preceding frame (the immediately preceding inter frame or the immediately preceding intra frame) input to the apparatus 1. , P), based on the result (residual signal) obtained by subtracting the local decoded image B ′ (c, p) output from the video compression unit 5, compressed data whose luminance value is less than the set value (minimum) is obtained. To choose.

  The value less than the set value is a value that does not disturb the luminance fluctuation. For example, in the case of an 8-bit image, it is 10 (this numerical value is an arbitrary number), and may be a sufficiently small value. However, if this set value is too large, flicker interference cannot be suppressed.

Further, the selection means 7 decodes the intra block image IB ′ immediately before decoding obtained by decoding the immediately preceding intra block image in the same area as the predetermined area in the intra frame immediately before the block image, instead of the block image B ′ (r, p) immediately before decoding. It is also possible to select compressed data that minimizes the luminance difference between (c _(n−1) , p) and the local decoded image B ′ (c _(n) , p). Here, c _(n) indicates the current intra frame, and c _(n-1) indicates the previous intra frame.

In other words, the prediction information / quantized value control compression encoding apparatus 1 uses the selection unit 7 to select the previous intra-frame at the time of decoding instead of the block image in the same area as the predetermined area of the previous frame when the bit rate is low. Luminance difference between the intra-block image IB ′ (c _(n−1) , p) immediately before decoding and the local decoded image B ′ (c _(n) , p) obtained by decoding the previous intra-block image in the same area as the predetermined area in the frame By selecting the compressed data that minimizes the flicker interference when the optimum compressed data output from the output means 9 is decoded.

Further, the selection means 7 can select compressed data based on a difference in hue in addition to a difference in luminance value. That is, the prediction information / quantization value control compression encoding apparatus 1 selects the compressed data based on the difference in hue as well as the difference in luminance value by the selection unit 7, that is, the block image B ′ immediately before decoding. When the optimum compressed data output from the output means 9 is decoded by selecting the compressed data that minimizes the hue difference between (r, p) and the local decoded image B ′ (c _(n) , p) In addition, flicker interference can be reduced.

  The output unit 9 includes a nonvolatile memory and has a buffering function, and holds a combination (compressed data) of the plurality of prediction information output from the video compression unit 5 and the corresponding quantized value. Thus, the compressed data selected by the selection means 7 is output as optimum compressed data.

According to the prediction information and quantization value control compression encoding apparatus 1, the video compression unit 5, the prediction information based on the reference image and the block image is estimated, a combination generating unit 3, based on the prediction information more multiple prediction information is generated. Then, in the image compression unit 5, on the basis of a plurality prediction information, the prediction signal is generated, the block image B (c, p) quantized values ized converting quantum subtraction signal obtained by subtracting the prediction signal from the Ru is output. Further, the combination generating unit 3 generates a combination of the plurality of prediction information and the corresponding quantized value, and the video compression unit 5 outputs compressed data obtained by compressing the combination, and the corresponding quantum in the compressed data. A local decoded image B ′ (c, p) in which the digitized values are combined is output. Then, the selection unit 7 compresses the optimum compressed data so that the compressed data in which the luminance difference between the block image B ′ (r, p) immediately before decoding and the local decoded image B ′ (c, p) is less than a preset set value And output by the output means 9. For this reason, flicker interference (image quality degradation) caused by a slight luminance difference between the block image B (c, p) and the immediately preceding block image B (r, p) in the immediately preceding frame being expanded by the quantization process This can be reduced by selecting a combination of multiple prediction information and a corresponding quantized value for which the luminance difference is less than the setting. In other words, flicker interference can be prevented in the compression encoding method using the prediction signal.

  Further, according to the prediction information / quantized value control compression encoding apparatus 1, flicker interference (image quality degradation) caused by a slight luminance difference between the block image and the immediately preceding block image being expanded by the quantization process is selected by the selection unit 7. By selecting a combination of prediction information and a quantized value that minimizes the luminance difference, it can be reduced.

<Operation of Prediction Information / Quantization Value Control Compression Encoding Device (First Embodiment)>
Next, the operation of the prediction information / quantized value control compression encoding apparatus 1 shown in FIG. 1 will be described with reference to the flowchart shown in FIG. 6 (see FIG. 1 as appropriate). The description of the operation of the prediction information / quantization value control compression encoding apparatus 1 shown in this flowchart is synonymous with the description of the prediction information / quantization value control compression encoding program operating on the computer.

  First, the prediction information / quantization value control compression encoding apparatus 1 minimizes the luminance difference between the local decoded image (decoded block image) B ′ (c, p) and the block image B ′ (r, p) immediately before decoding. A variable indicating a value (hereinafter referred to as a minimum value variable) is initialized (step S1). Note that the value to be initialized is larger than the maximum value that can be taken by the difference between the two block images (local decoded image B ′ (c, p) and block image B ′ (r, p) immediately before decoding).

Subsequently, the prediction information / quantized value control compression encoding apparatus 1 inputs the prediction information from the video compression means 5 to the combination generation means 3, that is, obtains prediction information from the combination generation means 3 (step S2). ). Then, the prediction information / quantized value control compression coding apparatus 1 determines a combination of prediction information by the combination generation unit 3, that is, generates a plurality of pieces of prediction information (step S3).

  Then, the prediction information / quantized value control compression encoding apparatus 1 outputs a plurality of pieces of prediction information from the combination generation unit 3 (step S4), and outputs a quantized value by the video compression unit 5, that is, the combination generation unit 3 Therefore, a quantized value is acquired (step S5). Then, the prediction information / quantized value control compression coding apparatus 1 determines a combination of corresponding quantized values based on the acquired quantized values by the combination generating means 3, that is, generates corresponding quantized values. (Step S6). Then, the prediction information / quantized value control compression encoding apparatus 1 outputs one of the corresponding quantized values to the video compression unit 5 by the combination generation unit 3 (step S7).

  The prediction information / quantized value control compression encoding apparatus 1 acquires a decoded image (local decoded image B ′ (c, p)) by the video compression means 5 (step S8), and this decoded image (local decoded image B). '(C, p)) and the block image B' (r, p)) immediately before decoding are compared with the minimum value variable by the selection means 7, and this difference is smaller than the minimum value (minimum value variable). Is determined (step S9). If it is determined that this difference is smaller than the minimum value (minimum value variable) (step S9, Yes), the minimum value (minimum value variable) is updated (step S10), and the multiple prediction information and the corresponding quantized value are compressed. The compressed data is saved (held) by the output means 9 (step S11).

  In step S9, if the difference between the decoded image (local decoded image B ′ (c, p)) and the block image B ′ (r, p) immediately before decoding is smaller than the minimum value (minimum value variable). If not determined (step S9, No), the prediction information / quantized value control compression encoding apparatus 1 determines whether or not there is a corresponding quantized value by the video compression unit 5 (step S12). When it is determined that there are still quantized values (step S12, Yes), the operations from step S7 to step S12 are repeated.

  Then, when the prediction information / quantization value control compression encoding apparatus 1 does not determine that the corresponding quantization value still exists by the video compression unit 5 (No in step S12), there is still multiple prediction information. Is determined (step S13). If it is determined that there is still multiple prediction information (step S13, Yes), the operations from step S4 to step S12 are repeated. Furthermore, when it is not determined that there is still multiple prediction information (No in step S13), the compressed data (the compressed multiple prediction information and the corresponding quantized value) stored (held) in the output unit 9 is optimized. Output as compressed data (step S14).

<Example of suppressing flicker interference>
Next, an example in which flicker interference is suppressed when a moving image is compression encoded by the prediction information / quantization value control compression encoding device 1 (1A, 1B) will be described with reference to FIGS. (See FIGS. 1 and 6 as appropriate).

  In FIGS. 7 and 8, in order to simplify the description, a prediction signal and prediction information are synonymous using a one-dimensional signal (original signal) that changes with time instead of a moving image, and encoding means 15 The conversion process in the conversion means 15a is omitted.

  FIG. 7A shows the original signal and the prediction signal. Note that the original signal in FIG. 7A is the same as the input signal (digital signal) input to the conventional flicker reduction circuit described in the background art. In FIG. 7A, description will be given focusing on the second sample (a signal between the first and second segments on the time axis).

  (A) shown to Fig.7 (a) is the prediction information (prediction signal) acquired by step S2 demonstrated by operation | movement of the said apparatus 1. FIG. In addition to (A), it is assumed that the prediction signals indicated by dotted lines in (A) and (C) were also tested (handled by the video compression means 5). The loop processing from step S4 to step S13 is performed for these three types (a), (b), and (c).

First, the case of the prediction signal (a) will be described.
The prediction signal (A) is output in step S4, which is the operation of the prediction information / quantization value control compression encoding apparatus 1. Here, FIG. 7B shows a residual signal obtained by subtracting the prediction signal from the original signal. Since the conversion process is omitted, a value obtained by quantizing the residual signal is an output from the video compression unit 5. FIG. 7C shows an example of the corresponding quantization value obtained from the video compression unit 5 in step S5 which is the operation of the apparatus 1.

  In step S6, three combinations of the obtained quantized value and a value obtained by adding ± 1 to the quantized value (corresponding quantized value) are determined. FIG. 7D shows the three corresponding quantization values determined in step S6. For these three corresponding quantized values, a loop process from step S7 to step 12 is performed. FIG. 7E shows the decoded signal (local decoded image B ′ (c, p)) obtained in step S8. Since the conversion process is omitted, in FIG. 7E, a value obtained by adding the prediction signal (a) shown in FIG. 7A to the corresponding quantized value shown in FIG. 7D is obtained.

  Next, in step S9 which is the operation of the apparatus 1, the difference between the decoded signal and the decoded signal in the immediately preceding frame (block image B ′ (r, p) immediately before decoding) is compared with the value of the minimum value variable. If this difference is smaller than the value of the minimum value variable, in step S10, the minimum value variable is set to the difference between the decoded signal and the decoded signal in the immediately preceding frame (block image B ′ (r, p) immediately before decoding). In step S11, the plurality of pieces of prediction information and the corresponding quantized values are stored.

  In the case of the prediction signal (a), both the case where the corresponding quantization value is -1 and the case where it is 0 are minimum values. As described above, when there are a plurality of signals in which the difference between the decoded signal and the decoded signal in the immediately preceding frame is minimum, the one with the better compression rate or the image quality of the decoded signal (decoded image) (in this case, corresponding quantization) Suppose that the value is 0).

  The above processing is also performed for the prediction signal (A) and the prediction signal (C). When the prediction signal (A) or the prediction signal (C) is used, the residual signal is as shown in FIG. 7B, and the quantized value obtained in step S5 which is the operation of the apparatus 1 is As shown in FIG.

  As in the case of the prediction signal (a), the corresponding quantized values determined in step S6 are shown in FIG. 8A (prediction signal (A)) and FIG. 8C (prediction signal (C)). The decoded signals obtained by decoding the corresponding quantized values are as shown in FIG. 8B (predicted signal (A)) and FIG. 8D (predicted signal (C)).

  The nine decoded signals shown in FIG. 7E, FIG. 8B, and FIG. 8D are all illustrated combinations. Therefore, in step S10, which is the operation of the apparatus 1, the decoded signal that minimizes the difference from the decoded signal of the immediately preceding frame is selected when the predicted signal (c) is selected and the corresponding quantized value is -1. is there. The combination of the prediction signal and the corresponding quantized value is output in step S14.

<Configuration of Prediction Information / Quantization Value Control Compression Encoding Device (Second Embodiment)>
FIG. 3 is a block diagram of a prediction information / quantization value control compression encoding apparatus (second embodiment). As shown in FIG. 3, the prediction information / quantization value control compression coding apparatus 1A compresses and encodes a moving image by controlling a combination of multiple prediction information and corresponding quantization values. Means 3, video compression means 5A, selection means 7, and output means 9 are provided. In addition, about the same structure as the structure of the prediction information and quantization value control compression encoding apparatus 1 shown in FIG. 1, the same code | symbol is attached | subjected and the description is abbreviate | omitted.

The combination generating unit 3A generates and outputs a plurality of pieces of prediction information whose number is limited in advance without inputting the prediction information from the video compression unit 5A, and based on the quantization value input from the video compression unit 5A. The corresponding quantized value is output. The number of combinations of multiple prediction information whose number is limited in advance and the corresponding quantized value is, for example, H.264. In the case of intra prediction in the H.246 system, there are 12 types (3 [three types of quantization values of −1, 0, +1] × 4 [number of intra 16 × 16 prediction modes in the H.264 system]). The number of combinations with the corresponding quantized values of the multiple prediction information is set to an appropriate number in consideration of the effect of suppressing flicker interference and the amount of calculation and balance (performance) of the apparatus 1A.

  The video compression unit 5A receives the block image B (c, p) to be compressed and the reference image, outputs the quantization value and the multiple prediction information and the corresponding quantization value between the combination generation unit 3A. And the compressed data and the local decoded image B ′ (c, p) are output.

  The video compression unit 5A will be described in detail with reference to FIG. 4 which is a block diagram in which the video compression unit 5A is developed (the configuration is divided). As shown in FIG. 4, the video compression unit 5 </ b> A includes a prediction signal generation unit 13, an encoding unit 15, a compressed data generation unit 17, and a decoding unit 19. In addition, about the same structure as the structure of the video compression means 5 shown in FIG. 2, the same code | symbol is attached | subjected and the description is abbreviate | omitted.

  As shown in FIG. 4, the video compression unit 5A is obtained by omitting the prediction information estimation unit 11 from the video compression unit 5 shown in FIG. 2, and the video compression unit 5A does not estimate the prediction information. Based on the multiple prediction information generated and output by the combination generation means 3A, the prediction signal generation means 13 generates a prediction signal.

According to the prediction information and quantization value control compression encoding device 1A, a plurality of multiple prediction information is generated by combining generator 3A, the video compression unit 5A, on the basis of a plurality prediction information generated prediction signal is, block image B (c, p) quantized values ized converting quantum subtraction signal obtained by subtracting the prediction signal from the Ru is output. Further, the combination generating unit 3A generates a combination of the plurality of prediction information and the corresponding quantized value, and the video compression unit 5A outputs compressed data obtained by compressing the combination, and the corresponding quantum in the compressed data. A local decoded image B ′ (c, p) in which the digitized values are combined is output. That is, since the plurality of pieces of prediction information is output by the combination generation unit 3A without estimating the prediction information by the video compression unit 5A, when the plurality of pieces of prediction information is limited, that is, a mode for generating a prediction signal, This can be applied when the number of values is limited, and, similarly to the prediction information / quantized value control compression encoding apparatus 1, flicker interference can be prevented in the compression encoding method using the prediction signal.

<Configuration of Prediction Information / Quantization Value Control Compression Encoding Device (Third Embodiment)>
FIG. 5 is a block diagram of a prediction information / quantization value control compression encoding apparatus (third embodiment). As shown in FIG. 5, the prediction information / quantization value control compression encoding apparatus 1B is configured to block a block image B (c, p) to be compression encoded and a block image B ′ (r, p) immediately before decoding. After the comparison, the combination of the plurality of prediction information and the corresponding quantized value is controlled to compress and encode the moving image. The combination generating means 3B, the video compressing means 5, the selecting means 7, and the output Means 9 and block comparison means 21 are provided. In addition, about the same structure as the structure of the prediction information and quantization value control compression encoding apparatus 1 shown in FIG. 1, the same code | symbol is attached | subjected and the description is abbreviate | omitted.

  Based on the comparison result output from the block comparison unit 21, the combination generation unit 3 </ b> B outputs only one combination of the multiple prediction information output to the video compression unit 5 and the corresponding quantized value. . That is, the combination generation unit 3B outputs the combination of the prediction information output from the video compression unit 5 and the quantized value as it is.

The block comparison unit 21 outputs a comparison result obtained by performing block comparison between the block image B (c, p) to be compression-encoded and the block image B ′ (r, p) immediately before decoding to the combination generation unit 3B. is there. The block comparison means 21 compares the two blocks (the block image B (c, p) and the block image B ′ (r, p) immediately before decoding) with the absolute value of the difference between the pixel values. Seeking in the sum. Note that the comparison of two blocks (difference magnitude) may be obtained by the sum of squares of differences between pixel values or the like. When this difference (for example, in the case of an 8-bit image, 30 [this numerical value is an arbitrary number]) is large enough to be recognized as a variation in luminance, the combination generating unit 3B combines the multiple prediction information and the corresponding quantized value. Are limited to one and output to the video compression means 5. Incidentally, it is sufficient that this difference is sufficiently large. On the contrary, if the difference is too small, the number of applicable blocks is reduced, and the effect of suppressing flicker interference is reduced.

  According to the prediction information / quantized value control compression encoding apparatus 1B, the block comparison unit 21 compares the block image B (c, p) with the block image B ′ (r, p) immediately before decoding, and compares them. Based on the comparison result, the video compression means 5 has a constant luminance difference between the block image B (c, p) and the block image B ′ (r, p) immediately before decoding according to the multiple prediction information output from the combination generation means 3B. It is also possible to process only the following block images. That is, the video compression unit 5 processes only the block image whose luminance difference from the block image immediately before decoding is a predetermined value or less, thereby reducing flicker interference when decoding the optimum compressed data output from the output unit 9. can do. That is, when the luminance difference from the block image immediately before decoding is larger than a certain value, it is assumed that there is a luminance variation in the original moving image (original image). There is no need to perform processing to reduce the interference, and the processing speed can be improved.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to the said embodiment. For example, in the present embodiment, the prediction information / quantization value control compression encoding apparatus 1 (1A, 1B) has been described. However, the processing of each component of the apparatus 1 (1A, 1B) is a general-purpose or special computer language. It can also be regarded as the prediction information / quantized value control compression encoding program described in (1). Moreover, it is also possible to regard the processing of each component of the apparatus 1 (1A, 1B) as a prediction information / quantized value control compression encoding method in which each process is regarded as an information processing process. In these cases, the same effect as that of the device 1 (1A, 1B) can be obtained.

1 is a block diagram of a prediction information / quantized value control compression coding apparatus (first embodiment) according to an embodiment of the present invention. FIG. It is the block diagram which expand | deployed the video compression means in FIG. It is a block diagram of the prediction information and quantization value control compression encoding apparatus (2nd embodiment) which concerns on embodiment of this invention. It is the block diagram which expand | deployed the video compression means in FIG. It is a block diagram of the prediction information and quantization value control compression encoding apparatus (3rd embodiment) which concerns on embodiment of this invention. It is the flowchart explaining operation | movement of the prediction information and quantization value control compression encoding apparatus (1st embodiment) shown in FIG. It is the figure explaining the example which suppressed flicker interference when the moving image was compression-encoded by the prediction information / quantized value control compression encoding device. FIG. 8 is a diagram for explaining an example in which flicker interference is suppressed when a moving image is compression-encoded by a prediction information / quantized value control compression encoding device (continuation of FIG. 7). It is the figure explaining the signal (signal waveform) after quantization which quantized the input signal (digital signal) in the conventional flicker reduction circuit (what demonstrated the view of a graph). It is a figure explaining the signal (signal waveform) after the quantization which quantized the input signal (digital signal) in the conventional flicker reduction circuit (the signal which showed the waveform after an input signal and quantization). It is a figure explaining the signal (signal waveform) after quantization at the time of applying the conventional flicker reduction circuit to the compression coding system using a prediction signal.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 1A, 1B Prediction information / quantization value control compression encoding apparatus 3, 3A, 3B Combination generation means 5, 5A Video compression means 7 Selection means 9 Output means 11 Prediction information estimation means 13 Prediction signal generation means 15 Encoding means (Quantized value output means)
15a Conversion means 15b Quantization means 17 Compressed data generation means (compressed data output means)
19 Decoding means (local decoded image output means)
19a Inverse quantization means 19b Inverse transformation means 21 Block comparison means

Claims (9)

Multiple prediction information generated based on prediction information for predicting an image of a predetermined block included in a moving image from a motion of the predetermined block or a region in the vicinity of the predetermined block, and the moving image are quantized A predictive information / quantized value control compression encoding apparatus that compresses and encodes the moving image by controlling a combination of the quantized value and a corresponding quantized value corresponding to the multiple prediction information,
The prediction including a type of prediction method and a value used for prediction based on a reference image that is a frame to be referred to when compression-encoding and a block image that is a block in a predetermined area of the frame to be compression-encoded Prediction information estimation means for estimating and outputting information;
Together with the generating a plurality of said plurality prediction information to the prediction information, and the plurality prediction information, and combining means for generating a combination of said corresponding quantized values corresponding to the plurality prediction information,
Based on the multiple prediction information generated by the combination generation means, a prediction signal generation means for generating a block of a predetermined region in the reference image as a prediction signal;
A quantized value output means for outputting a subtracted signal obtained by subtracting the predicted signal generated by the predicted signal generating means from the block image as a quantized value obtained by transform quantization ;
Compressed data generating means for compressing the combination of the multiple prediction information generated by the combination generating means and the corresponding quantized value, and outputting the compressed data as compressed data;
Decoding the corresponding quantized value included in the compressed data output by the compressed data generating means , generating a decoded signal by inverse quantization and inverse transform, and generating the decoded signal and the corresponding quantum A local decoded image output means for adding the prediction signal generated based on the plurality of prediction information to be combined with the quantization value and outputting as a local decoded image;
A setting value in which a luminance difference between a block image immediately before decoding the block image immediately before the predetermined area in the frame immediately before the block image and the local decoded image output by the local decoded image output unit is set in advance Selecting means for selecting the compressed data to be less than, and outputting selection information ;
And output means for outputting holds the compressed data output by the compressed data output means, said compressed data selected by the selection information in which the selection means is output as optimum compressed data,
With
The combination generating means includes
When the prediction information output by the prediction information estimation unit is input, based on the type of the prediction method, a plurality of predetermined values set in advance are added to values used for prediction included in the prediction information And generating a plurality of the plurality of prediction information,
When the quantized value output from the quantized value output unit corresponding to the generated multiple prediction information is input, a plurality of predetermined values set in advance are added to the quantized value component. Generating a plurality of corresponding quantized values, and generating a combination of the plurality of prediction information and the corresponding quantized value corresponding to the plurality of prediction information . apparatus. Predicting a predetermined block image included in a moving image from the movement of the predetermined block or the movement of an area in the vicinity of the predetermined block, a plurality of pieces of prediction information, and the moving image are quantized A predictive information / quantized value control compression encoding apparatus that compresses and encodes the moving image by controlling a combination of the quantized value and a corresponding quantized value corresponding to the multiple prediction information,
Thereby generating a plurality of said plurality prediction information including a value to be used in predicting the type of prediction methods, the plurality prediction information, the combination of the corresponding quantization values corresponding to the plurality prediction information generating A combination generating means for
Based on the multiple prediction information generated by the combination generation means, a prediction signal generation means for generating a block of a predetermined region in a reference image, which is a frame to be referred to when performing compression encoding, as a prediction signal;
A quantized value output means for outputting a subtracted signal obtained by subtracting the predicted signal generated by the predicted signal generating means from a block image that is a block in a predetermined region of a frame to be compressed and encoded as a quantized value obtained by transform quantization ;
Compressed data generating means for compressing the combination of the multiple prediction information generated by the combination generating means and the corresponding quantized value, and outputting the compressed data as compressed data;
Decoding the corresponding quantized value included in the compressed data output by the compressed data generating means , generating a decoded signal by inverse quantization and inverse transform, and generating the decoded signal and the corresponding quantum A local decoded image output means for adding the prediction signal generated based on the plurality of prediction information to be combined with the quantization value and outputting as a local decoded image;
A setting value in which a luminance difference between a block image immediately before decoding the block image immediately before the predetermined area in the frame immediately before the block image and the local decoded image output by the local decoded image output unit is set in advance Selecting means for selecting the compressed data to be less than, and outputting selection information ;
And output means for outputting holds the compressed data output by the compressed data output means, said compressed data selected by the selection information in which the selection means is output as optimum compressed data,
With
The combination generating means includes
When the quantized value output from the quantized value output unit corresponding to the generated multiple prediction information is input, a plurality of predetermined values set in advance are added to the quantized value component. Generating a plurality of corresponding quantized values, and generating a combination of the plurality of prediction information and the corresponding quantized value corresponding to the plurality of prediction information . apparatus.   The prediction information / quantized value control compression encoding apparatus according to claim 1, wherein the selection unit selects the compressed data that minimizes the luminance difference. The combination generating means includes
When the block image is not a block that performs intra prediction, when the prediction information output by the prediction information estimation unit is input, only the prediction information is output as the multiple prediction information, and the quantization when the value output means enter the quantized values output prediction information and quantization value control compressed codes according to claim 1, characterized in that it outputs only the quantized value as the corresponding quantized value Device. The combination generating means includes
When the block image is not a block in an intra frame, when the prediction information output by the prediction information estimation unit is input, only the prediction information is output as the multiple prediction information, and the quantization when the value output means enter the quantized values output prediction information and quantization value control compressed codes according to claim 1, characterized in that it outputs only the quantized value as the corresponding quantized value Device. Comparing the luminance difference between the block image and the block image immediately before decoding, and comprising block comparison means for outputting a comparison result,
The combination generating means includes
When the luminance difference included in the comparison result output by the block comparison unit is larger than a predetermined value, when the prediction information output by the prediction information estimation unit is input, the prediction information as the multiple prediction information outputs only, and, when inputting the quantized values the quantized value output means has output claim 1, characterized in that outputs only the quantized value as the corresponding quantized value Prediction information / quantization value control compression encoding apparatus.   The selection means includes, instead of the block image immediately before decoding, the luminance between the intra block image immediately before decoding and the local decoded image obtained by decoding the immediately preceding intra block image in the same region as the predetermined region in the intra frame immediately before the block image. 7. The prediction information / quantized value controlled compression encoding apparatus according to claim 1, wherein compressed data that minimizes the difference is selected.   8. The prediction information / quantized value according to claim 1, wherein the selection unit selects the compressed data based on a difference in hue in addition to the luminance difference. 9. Control compression encoding device. Multiple prediction information generated based on prediction information for predicting an image of a predetermined block included in a moving image from a motion of the predetermined block or a region in the vicinity of the predetermined block, and the moving image are quantized In order to compress and encode the moving image by controlling a combination with a corresponding quantization value corresponding to the multiple prediction information that is a quantization value,
The prediction including a type of prediction method and a value used for prediction based on a reference image that is a frame to be referred to when compression-encoding and a block image that is a block in a predetermined area of the frame to be compression-encoded Prediction information estimation means for estimating and outputting information,
Wherein together to generate a plurality of said plurality prediction information to the prediction information, the plurality prediction information and, combining means for generating a combination of said corresponding quantized values corresponding to the plurality prediction information,
Prediction signal generation means for generating a block of a predetermined region in the reference image as a prediction signal based on the multiple prediction information generated by the combination generation means,
A quantized value output means for outputting a subtracted signal obtained by subtracting the predicted signal generated by the predicted signal generating means from the block image as a quantized value obtained by transform quantization ;
Compressed data generating means for compressing the combination of the multiple prediction information generated by the combination generating means and the corresponding quantized value, and outputting the compressed data as compressed data;
Decoding the corresponding quantized value included in the compressed data output by the compressed data generating means , generating a decoded signal by inverse quantization and inverse transform, and generating the decoded signal and the corresponding quantum A local decoded image output means for adding the prediction signal generated based on the plurality of prediction information to be combined with the quantization value and outputting as a local decoded image;
A setting value in which a luminance difference between a block image immediately before decoding the block image immediately before the predetermined area in the frame immediately before the block image and the local decoded image output by the local decoded image output unit is set in advance Selecting means for selecting the compressed data to be less than, and outputting selection information ;
Output means for outputting holds the compressed data output by the compressed data output means, said compressed data selected by the selection information in which the selection means is output as optimum compressed data,
To function as,
The combination generating means includes
When the prediction information output by the prediction information estimation unit is input, based on the type of the prediction method, a plurality of predetermined values set in advance are added to values used for prediction included in the prediction information And generating a plurality of the plurality of prediction information,
When the quantized value output from the quantized value output unit corresponding to the generated multiple prediction information is input, a plurality of predetermined values set in advance are added to the quantized value component. Generating a plurality of corresponding quantized values, and generating a combination of the plurality of prediction information and the corresponding quantized value corresponding to the plurality of prediction information . program.

Images