JP2016082395A - Encoder, coding method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow for accurate control of the amount of generated codes, while suppressing the storage capacity.SOLUTION: When the difference absolute value of the target code amount related to the block line one before and the amount of generated codes goes above a predetermined threshold, first quantization step is set at first, and the weight of blocks included in an objective block line is determined, using the complexity of each block included in the block line one before. A second quantization step of each block is determined by integrating the weight thus determined in the first quantization step.SELECTED DRAWING: Figure 1

Description

  In particular, the present invention relates to an encoding apparatus, an encoding method, and a program suitable for use in controlling the amount of generated code.

  In recent years, a method of recording a RAW image has been applied to not only a still image but also a moving image. A RAW image has a large amount of data necessary for recording, while correction and deterioration of the original image can be minimized, so that the degree of freedom of image editing after shooting is high. For this reason, RAW images are favorably used by advanced users among those who use the imaging apparatus. However, when recording a moving image of a RAW image, compression encoding is required to compress the data amount to a desired code amount so that the moving image can be recorded for a predetermined time on a predetermined recording medium.

  As a conventional compression encoding method for moving images, H.264 is available. H.264 (H.264 / MPEG-4 Part 10: Advanced Video Coding) is known. In such a compression encoding method, the data amount is compressed using temporal redundancy and spatial redundancy of a moving image for each block having a predetermined number of pixels in one frame. H. In H.264, compression encoding is realized by combining encoding methods such as motion detection and motion compensation for temporal redundancy, discrete cosine transform and quantization for spatial redundancy, and further entropy encoding.

  As such a compression encoding method for each block, Patent Document 1 discloses a compression encoding method in which compression encoding is performed in units of blocks and code amount control is performed for each block line including one block. . According to the method described in Patent Document 1, the generated code amount is measured for each block line, and the code amount control is performed for each block line so that the generated code amount of one entire screen is equal to or less than a predetermined target code amount. . As a result, compression encoding can be performed to a desired code amount.

  Further, in Patent Document 2, after determining a quantization step, which is a quantization parameter, for each block, correction for the quantization step is performed in a direction to reduce the degree of deviation of the generated code amount from the target code amount. A compression coding control system that controls the above is disclosed. According to the method described in Patent Document 2, compression encoding is performed for each block, and the code amount control of a block to be encoded next is performed using the already encoded block, thereby compressing to a desired code amount. It can be encoded.

Japanese Patent Laid-Open No. 4-18857 JP 2002-94989 A

  By the way, with the recent evolution of image sensors, the number of pixels per image has increased significantly, and the number of images that can be continuously shot per second has also increased. For this reason, also in the compression encoding process, the ability to process such an input image efficiently and quickly is required. Furthermore, since the amount of image data handled increases enormously, the development processing also increases enormously, so that the tightness of the RAM bus bandwidth becomes a problem.

  In view of this, it is conceivable that the input image signal is not saved in the RAM, but is directly input in raster order from the sensor in units of pixel lines and is subjected to compression encoding, thereby reducing the RAM bandwidth. However, in compression encoding in units of pixel lines, the quantization step of each block must be determined when compression encoding of the first pixel line of each block starts. The code amount control using can not be performed.

  On the other hand, generally, a complex image has a large amount of generated code, and a simple image tends to have a small amount of generated code. For this reason, in the quantization process, which is one of the encoding processes, the amount of generated code is controlled by setting a large quantization step for complex images and a small quantization step for simple images. Can be considered. However, when a common quantization step is set for each block included in the block line as in the method described in Patent Document 1, in the case of an image having greatly different features on the left and right of the block line, a complicated region and a simple The area is compressed and encoded with a common quantization step. For this reason, in the case of an image having such characteristics, it is difficult to match the generated code amount to the desired code amount.

  The present invention has been made in view of the above-described problems, and it is an object of the present invention to control the amount of generated codes with high accuracy by reducing the storage capacity.

  An encoding apparatus according to the present invention is an encoding apparatus that encodes input image data in units of pixel lines, and is quantized by the quantization means for quantizing the image data. A comparison unit that compares the generated code amount up to the upper block line of the block to be processed and a target code amount in units of the pixel line, and a control that controls the quantization step in block units according to the comparison result by the comparison unit The control means determines a first quantization step of a block line to which a block to be quantized by the quantization means belongs according to a comparison result by the comparison means, and the quantization means A block unit by weighting the first quantization step based on information on the upper block of the block quantized by Determining a second quantization step, the quantization unit may be quantized by the second quantization step determined by the control means.

  According to the present invention, it is possible to accurately control the amount of generated code so as to reduce the storage capacity.

It is a block diagram which shows the structural example of the image coding apparatus which concerns on 1st Embodiment. It is a figure for demonstrating the input order of the image data input into an image coding part. It is a figure for demonstrating arrangement | positioning of a rectangular block. It is a flowchart which shows an example of the process sequence which controls the generated code amount by a quantization control part. It is a figure for demonstrating the acquisition range of the weighting information utilized when the 2nd block line is compression-encoded. It is a figure which shows the relationship between the change of the 2nd quantization step based on weighting information, and transition of code amount in a block line. It is a block diagram which shows the structural example of the image coding apparatus which concerns on 2nd Embodiment. It is a figure for demonstrating the subband formed when a discrete wavelet transform is each implemented twice in a horizontal direction and a perpendicular direction. It is a figure for demonstrating the process which determines the weighting information for every level paying attention to the level 1 of discrete wavelet transform. It is a figure which shows an example of the acquisition range of the weighting information utilized when a 2nd block line is compression-encoded in 4th Embodiment. It is a figure which shows the relationship between the change of the 2nd quantization step based on the weight adapted in 5th Embodiment, and transition of code amount.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram illustrating a configuration example of an image encoding device 10 according to the present embodiment. Hereinafter, the outline of the encoding process in the image encoding device 10 according to the present embodiment will be described with reference to FIG.
The image encoding unit 100 converts input image data for each pixel line in accordance with a quantization step which is a reference of an image set by the quantization setting unit 103 and a target code amount set by the target code amount setting unit 106. Are compressed and encoded data is output.

  FIG. 2 is a diagram for explaining the input order of the image data input to the image encoding unit 100 in FIG. As shown in FIG. 2, the image data is input to the image encoding unit 100 in a raster order in units of pixel lines, and is compressed and encoded. Note that the image data may be the pixel value itself input to the imaging device, or data obtained by performing some processing on the pixel value.

  FIG. 3 is a diagram for explaining the arrangement of rectangular blocks, and FIG. 3A is a diagram showing the pixel arrangement of one screen. FIG. 3B is a diagram showing an arrangement of rectangular blocks of one screen based on the pixel arrangement shown in FIG. As shown in FIG. 3, one image (one screen) is composed of n × m pixels, and in this embodiment, the code amount control is performed for each block with p × q pixels as one block. Note that, when an image input to the imaging apparatus is subjected to component decomposition and compression encoded for each component, the target code amount setting unit 106 in FIG. 1 described later sets a target code amount for each component. Therefore, one screen shown in FIG. 3 is also a unit for setting the target code amount by the target code amount setting unit 106.

  The quantization unit 101 quantizes the image data for each pixel in the quantization step set by the quantization control unit 110. The quantization step holding unit 102 holds the quantization step used when quantization is performed by the quantization unit 101 and the quantization step serving as a reference of the image set by the quantization setting unit 103. The encoding unit 104 entropy-encodes the image data quantized by the quantization unit 101 to generate encoded data, and holds the generated code amount in the generated code amount holding unit 105.

  The pixel line target code amount calculation unit 107 calculates a target code amount per pixel line using information stored in the quantization step holding unit 102, the generated code amount holding unit 105, and the target code amount setting unit 106. The pixel line target code amount holding unit 108 holds it. The code amount comparison unit 109, for each block line composed of one row of blocks shown in FIG. 3, stores the accumulated amount of generated code amounts up to the target block line and the pixel line target code amount held in the generated code amount holding unit 105. The integrated amount of the target code amount held in the holding unit 108 is compared. Then, the difference absolute value is notified to the quantization control unit 110.

  The quantization control unit 110 controls the quantization step for each block of image data so that the integrated amount of the generated code amount approaches the integrated amount of the target code amount.

  Specifically, first, the first quantization step, which is a block line quantization step, is determined from the comparison result (difference absolute value) of the previous block line calculated by the code amount comparison unit 109. Next, weighting information is calculated using the quantization step of each block included in the previous block line obtained from the information held in the quantization step holding unit 102 and the generated code amount holding unit 105, and the generated code amount. To do. Then, based on the calculated weighting information, the first quantization step is weighted to determine the second quantization step that is the quantization step for each block. The weighting information and the detailed contents of the weight will be described later. Thereby, the quantization unit 101 performs the quantization process using the second quantization step.

<Quantization control processing>
FIG. 4 is a flowchart illustrating an example of a processing procedure performed by the quantization control unit 110. Details of code amount control for each block in compression encoding in units of pixel lines will be described with reference to FIG. Hereinafter, (i) indicates an i-th block line, and (i, j) indicates a j-column block included in the i-th block line.

First, in step S <b> 401, the reference quantization step Q _ini for the one screen set by the quantization setting unit 103 is acquired from the quantization step holding unit 102. In step S402, it is determined whether the block line that controls the quantization step is the first line of the screen. As a result of this determination, if it is the first line, the process proceeds to S404; otherwise, the process proceeds to S403.

  In S403, it is determined whether or not the absolute value of the difference between the target code amount and the generated code amount related to the block line immediately before the block line, which is the comparison result of the code amount comparison unit 109, is smaller than a predetermined threshold value. To do. As a result of the determination, if it is less than the threshold value, the process proceeds to S404, and if not, the process proceeds to S405.

In S404, the reference quantization step Q _ini acquired in S401 is set as the first quantization step Q (i) of the block line. Then, the first quantization step is set as it is as the second quantization step Q (i, j) of each block included in the block line.

On the other hand, in S405, the first quantization step Q (i) of the block line is set as the reference quantization step Q _ini + α (i). Details of the value α will be described later. Next, in S406, the weight β (i, j) of the block included in the target block line using the weighting information I (i-1, j) that is the feature of each block included in the previous block line. ). Details of the weighting information I and the weighting β will also be described later.

  Subsequently, in S407, for each block, the weight β (i, j) determined in S406 is multiplied by the first quantization step Q (i) of the block line determined in S405. By this calculation, the second quantization step Q (i, j) for each block of the block line is determined. Note that the image data is input to the image encoding unit 100 in units of pixel lines. Therefore, the quantization unit 101 performs the quantization process using the second quantization step determined in S404 or S407 until all the blocks included in the block line are encoded by the encoding unit 104. become.

<First quantization step calculation method>
Next, a detailed calculation method of the value α added when calculating the first quantization step Q (i) in S405 will be described. As one of the methods for calculating the first quantization step, a known technique shown in MPEG2 Test Model 5 is known. According to the technique described in MPEG2 Test Model 5, as a method for controlling the code amount, a block line quantization step Q (i) to be controlled next is given by the following equation (1).
Q (i) = Q _ini + r × error (1)

Here, r represents a constant representing control sensitivity, and error represents the difference between the integrated amount of the generated generated code amount and the integrated amount of the target code amount. That is, when calculating the first quantization step Q (i), the technique described in MPEG2 Test Model 5 is applied, and the value α is defined by the following equation (2).
α = r × error (2)

<Second quantization step calculation method>
Next, a detailed calculation method of the weighting information I used in S405 and the weight β added in S406 will be described. First, in general, the features of images of adjacent blocks are similar, and the closer the coordinate position is, the higher the correlation is. Therefore, weighting information I (i−1, j) obtained for the entire block located immediately above the block is used as weighting information used for the target block.

  FIG. 5 is a diagram for explaining an acquisition range of weighting information used when the second block line is compression-encoded. As shown in FIG. 5, the weighting information used for the block (i, j) included in the second block line is represented by weighting information I (i-1, j) for the entire block located in the upper stage of the block. be able to.

  Next, in the present embodiment, image features are used as weighting information. When attention is paid to the generated code amount of an image, a block having a complicated image generates a larger code amount when the quantization process is performed in the same quantization step as compared with a block having a flat image. Therefore, the weighting information I (i−1, j) used for the block (i, j) uses the complexity X (i−1, j) of the image, and the complexity X (i−1, j). Is used to determine the weight β (i, j) of the block included in the block line.

That is, according to FIG. 5, the weighting information I (i−1, j) used for the block (i, j) is the complexity X () of the block (i−1, j) located immediately above the block. i-1, j). This complexity is characterized by being large if the image in the unit screen is complex and small if it is flat. The complexity of the block (i−1, j) is expressed by the following equation (2) using the second quantization step Q (i−1, j) and the generated code amount S (i−1, j) of the block ( 3).
X (i−1, j) = Q (i−1, j) × S (i−1, j) (3)

As described above, the weight β (i, j) is, for example, as shown in the following equation (4) with respect to the average value X (i−1) ave. Of the complexity of each block included in the previous block line. , And can be determined by the deviation (ratio) of the complexity of the block.
β (i, j) = X (i−1, j) / X (i−1) ave. (4)

  FIG. 6 is a diagram illustrating the relationship between the change in the second quantization step based on the weighting information and the change in the code amount in the block line. FIG. 6 shows an example in which the number of blocks in the block line is six. FIG. 6A is a diagram illustrating a change in the weighting information I (i−1, j) for each block in the previous block line. In the present embodiment, the weighting information I (i−1, j) in FIG. 6A is equivalent to the complexity X (i−1, j). FIG. 6B shows a change in the second quantization step Q (i, j) with respect to the first quantization step Q (i).

  FIG. 6C is a diagram showing the transition of the integrated amount of the generated code amount of the block line when quantization is performed in the second quantization step shown in FIG. 6B. In FIG. 6C, a dotted line 601 represents a change in the integrated amount of the target code amount in the target block line. A broken line 602 represents a change in the integrated amount of the generated code amount when the second quantization step is a constant value in each block of the block line. A point 603 represents the integrated amount of the generated code amount for each block when the second quantization step shown in FIG. 6B is applied to each block. The generated code amount of the block line is equivalent to the integrated amount of the generated code amount of the sixth block.

  As can be seen from the polygonal line 602, in the example shown in FIG. 6C, the left block of the block lines has a simpler pattern, so the amount of generated code in the block is smaller, and the right block has a relatively smaller pattern. Due to the complexity, the amount of generated code per block increases. If a uniform second quantization step is used in a block line having such characteristics, the generated code amount is likely to deviate significantly from the target code amount of the block line. On the other hand, if the second quantization step is changed for each block using the weighting information I (i−1, j) as in the present embodiment, the generated code amount of the block line is reduced as indicated by a point 603. It becomes possible to get closer to the target code amount.

  As described above, according to the present embodiment, even when image data is input in units of pixel lines, the quantization step for each block can be adaptively changed according to the characteristics of the image. It is possible to provide an image encoding device that is made higher.

(Second Embodiment)
Hereinafter, a second embodiment of the present invention will be described.
FIG. 7 is a block diagram illustrating a configuration example of the image encoding device 70 according to the present embodiment. In the present embodiment, it is assumed that frequency conversion is performed on image data. The same components as those in FIG. 1 are denoted by the same reference numerals, and description thereof is omitted. In the configuration illustrated in FIG. 7, the image encoding unit 700 includes a discrete wavelet transform unit 711 and a transform coefficient holding unit 712. Hereinafter, differences from the first embodiment will be described.

  The discrete wavelet transform unit 711 transforms the image data into a frequency domain signal (hereinafter, discrete wavelet transform), and outputs the transform coefficient to the quantization unit 701 and the transform coefficient holding unit 712. The quantization unit 701 quantizes the input transform coefficient in accordance with an instruction from the quantization control unit 710. The quantization control unit 710 calculates a quantization step using information held by the transform coefficient holding unit 712. Details of these processes will be described later.

  In compression coding, by performing compression coding of frequency transform coefficients, processing is performed to roughen the quantization step for images with high spatial frequency that are considered insensitive to human visual characteristics. Can do. For this reason, compression encoding processing can be performed more efficiently.

<Discrete wavelet transform>
FIG. 8 is a diagram for explaining subbands, which are partial frequency bands, formed when the discrete wavelet transform is performed twice each in the horizontal direction and in the vertical direction. In the discrete wavelet transform, the image is divided into four subbands LL, HL, LH, and HH by performing filtering once each in the horizontal direction and the vertical direction of the image. Here, L indicates a low-frequency component, H indicates a high-frequency component, and indicates horizontal and vertical bands in the order of characters. For example, HL represents a partial frequency band in which the horizontal component is a high frequency and the vertical component is a low frequency. In the discrete wavelet transform, such processing can be repeatedly performed on the LL component, and levels Lv corresponding to the number of times of the discrete wavelet transform are constructed. If the number of wavelet transforms is N, a total of N × 3 + 1 regions are formed for the subbands.

<Quantization control processing>
The processing procedure of the quantization control unit 710 in this embodiment is basically the same procedure as the flowchart of FIG. 4 described in the first embodiment, and controls one screen as one subband. However, in the present embodiment, the calculation method of the weighting information I and the weight β is different from that in the first embodiment. Hereinafter, a detailed calculation method of the weighting information I and the weight β will be described.

  Also in the present embodiment, as the weighting information I used for the target block, the weighting information I (i−1, j) obtained over the entire block located immediately above the block is used. The transform coefficient of the discrete wavelet transform held in the transform coefficient holding unit 712 is large for image data with low correlation among the partial frequency bands represented by subbands, and is small otherwise. Further, the larger the conversion coefficient, the larger the generated code amount. In other words, a block with a large total Coeff.sum of transform coefficients in a block has a larger code amount when quantized in the same quantization step than a block with a small total Coeff.sum of transform coefficients. It is expected to occur.

Therefore, in the present embodiment, the weighting information I (i−1, j) used for the target block includes the sum of transform coefficients for each block included in the previous block line Coeff.sum (i− 1, j). Using the transform coefficient, the value of the weight β is, for example, the average value Coeff. (I−1) of the sum of transform coefficients of each block included in the previous block line as in the following equation (5). It can be determined by the bias of Coeff.sum (i-1, j) with respect to ave.
β (i, j) = Coeff.sum (i−1, j) / Coeff. (i−1) ave. (5)

  The effect of the present embodiment according to the weight β calculation method described above will be described with reference to FIG. In the present embodiment, the weighting information I (i−1, j) in FIG. 6A is equivalent to the total sum of transform coefficients Coeff.sum (i−1, j). As described in the first embodiment, when the uniform second quantization step is used in the block line, the generated code amount is larger than the target code amount in an image having greatly different generated code amounts on the left and right of the block line. Is likely to come off. On the other hand, if the second quantization step is changed for each block using the weighting information I (i−1, j) as in this embodiment, the generated code amount can be made closer to the target code amount.

  As described above, according to the present embodiment, when image data is input in units of pixel lines, an image coding apparatus that performs discrete wavelet transform also uses the transform coefficient of discrete wavelet transform for each block. You can change the quantization step. Therefore, it is possible to provide an image encoding device that can adaptively change the quantization step for each block in accordance with the characteristics of the image and increase the accuracy of code amount control.

  Note that the weight β for each subband determined by the above weighting information may be further weighted in consideration of visual characteristics. For example, only the weight β of the subband HH having the highest spatial frequency can be increased with respect to the subbands HL and LH. In the present embodiment, it may be possible to switch whether or not the code amount control is performed depending on the subband.

(Third embodiment)
The configuration example of the image encoding device according to the present embodiment is the same as the configuration of FIG. 7 in the second embodiment. In the present embodiment, the quantization control unit 710 controls the second quantization step in units of levels using specific one weighting information among subbands having the same discrete wavelet transform level. Since the basic operation of the other configuration is the same as that of the second embodiment, description thereof is omitted.

  In the discrete wavelet transform, an image is divided in units of subbands that are partial frequency bands. In general, subbands at the same level in the discrete wavelet transform are closer in frequency band than subbands at different levels, and therefore, the degree of deterioration of the original image after decoding by quantization processing is also close. Therefore, an image obtained by performing quantization processing on subbands of the same level in completely different quantization steps and decoding from subbands having different deterioration levels tends to be subjectively unnatural images. From the above, it is desirable not to provide a certain difference in quantization step between subbands within the same level. Therefore, in the present embodiment, the same weighting information is used in three subbands within the same level, thereby preventing a difference in quantization step between subbands.

  FIG. 9 is a diagram for explaining processing for determining weighting information for each level, focusing on level 1 of the discrete wavelet transform. FIG. 9A is a diagram for explaining the flow of processing for determining weighting information, and FIG. 9B is a diagram showing an example of the sum of transform coefficients in the three subbands of level 1. is there.

<Quantization control processing>
Next, the code amount control method of this embodiment will be described with reference to FIG. In this embodiment, the processing procedure of the quantization control unit 710 is the same as that in the flowchart of FIG. 4, and this embodiment is different from the second embodiment in the weighting information selection method.

  First, in the present embodiment, after image data is subjected to discrete wavelet transform, compression encoding processing is sequentially performed for each pixel line. As shown in FIG. 9A, the transform coefficient holding unit 712 holds transform coefficients of discrete wavelet transform of each subband. As the weighting information I (i−1, j) used for the target block line, the quantization control unit 710 has the sum of transform coefficients in the previous block line Coeff. (I−1) being the most within the level. Coeff.sum (i-1, j) of a large subband is used.

In the case of the example shown in FIG. 9B, since the sum of the LH component transform coefficients Coeff. (I−1) is the largest among the subbands at the same level, the quantization control unit 710 weights the LH component. Information is used for quantization control of all subbands of Lv1. That is, the value of the weight β is, for example, for the average value Coeff (i−1) .ave. Of the total sum of transform coefficients of each block included in the previous block line as in the following equation (6). It can be determined by the bias of Coeff.sum (i-1, j).
β (i, j) _Lv1_HL
= Β (i, j) _Lv1_LH
= Β (i, j) _Lv1_HH
= Coeff.sum (i-1, j) _Lv1_LH / Coeff. (I-1) ave._Lv1_LH
... (6)

  As described above, according to the present embodiment, since the same weighting information is used between subbands at the same level, it is possible to reduce the difference in quantization step between subbands at the same level. Accordingly, it is possible to provide an image encoding device that increases the accuracy of code amount control while preventing a decoded image from becoming a visually unnatural image.

(Fourth embodiment)
The configuration example of the image encoding device according to the present embodiment is the same as the configuration of FIG. 1 in the first embodiment. In the present embodiment, the quantization control unit 110 calculates the weight β used for the block using the weighting information included in the pixel line of the lower arbitrary row among the blocks directly above the target block. Since the basic operation of the other configuration is the same as that of the first embodiment, description thereof is omitted.

  In general, since the similarity of regions in the image where the coordinate position is close is high, the weighting information used to calculate the weight is also more reliable as the coordinate is closer to the target block. To do. Therefore, the weighting information acquisition range may not be the entire block located immediately above the block.

  FIG. 10 is a diagram illustrating an example of an acquisition range of weighting information used when the second block line is compression-encoded. FIG. 10 shows that the information used when calculating the weight β for each block is an arbitrary number of lower pixel lines among the blocks located in the upper stage of the block. When the number of vertical pixels of the block is large, weighting information can be obtained from images with higher similarity by limiting the number of pixel lines to be referred to to a part as in this embodiment.

  As described above, according to the present embodiment, the weighting information used to determine the second quantization step for each block is lower than the block located immediately above the block that is closer to the block. Obtain from any number of pixel lines. As a result, it is possible to provide an image encoding device that can obtain weighting information from images with high similarity and that has higher accuracy in code amount control.

(Fifth embodiment)
The configuration example of the image encoding device according to the present embodiment is the same as the configuration of FIG. 1 in the first embodiment. In the present embodiment, the quantization control unit 110 determines the weight for calculating the second quantization step depending on the magnitude of the predetermined threshold. Since the basic operation of the other configuration is the same as that of the first embodiment, description thereof is omitted.

  FIG. 11 is a diagram illustrating the relationship between the change in the second quantization step and the change in the code amount based on the weights applied in the present embodiment. FIG. 11 shows an example in which the number of blocks in the block line is six. FIG. 11A is a diagram illustrating changes in the weighting information I (i−1, j) for each block in the previous block line. As shown in FIG. 11A, a first weighting threshold Th1 and a second weighting threshold Th2 are set for the weighting information. In this embodiment, the weighting information I (i−1, j) in FIG. 11A is equivalent to the complexity X (i−1, j).

  FIG. 11B is a diagram showing a change in the second quantization step Q (i, j) with respect to the first quantization step Q (i). FIG. 11C is a diagram showing the transition of the integrated amount of the generated code amount of the block line when quantization is performed in the second quantization step shown in FIG. 11B.

  In FIG. 11C, a dotted line 1101 represents a change in the integrated amount of the target code amount in the target block line. A broken line 1102 represents a change in the integrated amount of the generated code amount when the second quantization step is a constant value in each block of the block line. A curve 1103 represents a change in the integrated amount of the generated code amount for each block when the second quantization step shown in FIG. 11B is applied to each block. The generated code amount of the block line is equivalent to the integrated amount of the generated code amount of the sixth block. Moreover, in FIG.11 (c), the part in which the broken line 1102 and the curve 1103 are changing substantially parallel has shown that the quantization step is equal. In FIG. 11 (c), only blocks in which the second quantization step is changed are indicated by circles, and blocks that are not are indicated by ×.

  In the present embodiment, as shown in FIG. 11, two weighting thresholds are set when calculating the weight β for each block. The first weighting threshold Th1 is set to select a block whose image is very complicated. If the complexity is greater than the first weighting threshold Th1, a second quantization step greater than the first quantization step is provided. That is, the weight β (i, j) is set to the value β1 (β1> 1).

  On the other hand, the second weighting threshold Th2 is set to select a block whose image is very flat. If the complexity is smaller than the second weighting threshold Th2, a second quantization step smaller than the first quantization step is given. That is, the weight β (i, j) is set to the value β2 (β2 <1).

  Further, when the complexity of the block is not less than the second weighting threshold Th2 and not more than the first weighting threshold Th1, no weight is assigned. That is, the weight β (i, j) is set to 1.

  As shown in FIG. 11C, since the second quantization step other than the specific block in the block line remains the first quantization step, the variation in the quantization step can be reduced. In the code amount control, it is important to make the generated code amount close to the target code amount. On the other hand, if the quantization step for each block varies, it may be visually noticeable as a difference in image quality between the blocks. Therefore, by arbitrarily setting the upper limit of the value β1 and the lower limit of the value β2 and setting the second quantization step in stages, it is possible to further suppress deterioration in image quality between blocks.

  As described above, by applying the weight calculation method shown in this embodiment, it is possible to increase the accuracy of code amount control while reducing the difference in image quality. In other words, unless the weighting information is an extreme value, the first quantization step is adapted to the second quantization step, so that image coding with high accuracy of code amount control is avoided while avoiding visual image degradation. An apparatus can be provided.

(Other embodiments)
As mentioned above, although each embodiment was explained in full detail, this invention is not limited to a specific embodiment, A various deformation | transformation and change are possible. It is also possible to combine all or a plurality of the constituent elements of the above-described embodiment.

  The present invention supplies a program that realizes one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in a computer of the system or apparatus read and execute the program This process can be realized. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

101 Quantization unit 104 Encoding unit 109 Code amount comparison unit 110 Quantization control unit

Claims (13)

An encoding device that encodes input image data in units of pixel lines,
Quantization means for quantizing the image data;
Comparison means for comparing the generated code amount up to the upper block line of the block quantized by the quantization means with the target code amount in units of the pixel line;
Control means for controlling the quantization step in block units according to the comparison result by the comparison means,
The control means determines a first quantization step of a block line to which a block to be quantized by the quantization means belongs according to a comparison result by the comparison means, and a block to be quantized by the quantization means Determining a second quantization step in units of blocks by weighting the first quantization step based on information relating to the upper block of
The coding apparatus characterized in that the quantization means quantizes in a second quantization step determined by the control means.   The control means weights the first quantization step using a generated code amount and a quantization step related to an upper block of the block quantized by the quantization means. Item 4. The encoding device according to Item 1.   The control means is a block quantized by the quantization means for an average value of a product of a generated code amount and a quantization step related to a block belonging to an upper block line of the block quantized by the quantization means. The encoding apparatus according to claim 2, wherein the first quantization step is weighted by a ratio of a product of a generated code amount and a quantization step related to the upper block.   The control means gradually changes to the first quantization step according to the product of the generated code amount and the quantization step related to the block belonging to the upper block line of the block quantized by the quantization means. The encoding apparatus according to claim 1, wherein weighting is performed on the encoding apparatus.   5. The control unit according to claim 1, wherein the control unit weights the first quantization step based on information relating to the entire upper block of the block quantized by the quantization unit. The encoding device according to claim 1.   The control means weights the first quantization step based on information on a part of pixel lines in an upper block of the block quantized by the quantization means. The encoding apparatus of any one of Claims 1-4. An encoding device for encoding input image data in units of pixel lines by performing discrete wavelet transform,
Conversion means for converting the frequency of the input image data;
Quantization means for quantizing the transform coefficient of the frequency transform by the transform means;
Comparison means for comparing the generated code amount up to the upper block line of the block quantized by the quantization means with the target code amount in units of the pixel line;
Control means for controlling the quantization step in block units according to the comparison result by the comparison means,
The control means determines a first quantization step of a block line to which a block to be quantized by the quantization means belongs according to a comparison result by the comparison means, and a block to be quantized by the quantization means Determining a second quantization step in units of blocks by weighting the first quantization step based on a transform coefficient associated with the upper block of
The coding apparatus characterized in that the quantization means quantizes in a second quantization step determined by the control means.   The said control means weights with respect to said 1st quantization step using the sum total of the transformation coefficient which concerns on the block of the upper stage of the block quantized by the said quantization means. Encoding device.   8. The encoding apparatus according to claim 7, wherein the control unit weights the first quantization step according to the number of times frequency conversion is performed by the conversion unit. An encoding method for encoding input image data in units of pixel lines,
A quantization step of quantizing the image data;
A comparison step of comparing the generated code amount up to the upper block line of the block to be quantized in the quantization step with a target code amount in units of the pixel line;
A control step of controlling the quantization step in block units according to the comparison result in the comparison step,
In the control step, a first quantization step of a block line to which a block to be quantized in the quantization step belongs is determined according to a comparison result in the comparison step, and is quantized in the quantization step. Determining a second quantization step in block units by weighting the first quantization step based on information relating to the upper block of the block;
In the quantization step, quantization is performed by the second quantization step determined in the control step. An encoding method for encoding input image data by discrete wavelet transform and encoding in units of pixel lines,
A conversion step of converting the frequency of the input image data;
A quantization step of quantizing the conversion coefficient of the frequency conversion in the conversion step;
A comparison step of comparing the generated code amount up to the upper block line of the block to be quantized in the quantization step with a target code amount in units of the pixel line;
A control step of controlling the quantization step in block units according to the comparison result in the comparison step,
In the control step, a first quantization step of a block line to which a block to be quantized in the quantization step belongs is determined according to a comparison result in the comparison step, and is quantized in the quantization step. Determining a second quantization step in block units by weighting the first quantization step based on a transform coefficient associated with an upper block of the block;
In the quantization step, quantization is performed by the second quantization step determined in the control step. A program for controlling an encoding device that encodes input image data in units of pixel lines,
A quantization step of quantizing the image data;
A comparison step of comparing the generated code amount up to the upper block line of the block to be quantized in the quantization step with a target code amount in units of the pixel line;
A control step for controlling the quantization step in block units according to the comparison result in the comparison step is executed by a computer,
In the control step, a first quantization step of a block line to which a block to be quantized in the quantization step belongs is determined according to a comparison result in the comparison step, and is quantized in the quantization step. Determining a second quantization step in block units by weighting the first quantization step based on information relating to the upper block of the block;
In the quantization step, the program is quantized by the second quantization step determined in the control step. A program for controlling an encoding apparatus that encodes input image data in units of pixel lines by performing discrete wavelet transform,
A conversion step of converting the frequency of the input image data;
A quantization step of quantizing the conversion coefficient of the frequency conversion in the conversion step;
A comparison step of comparing the generated code amount up to the upper block line of the block to be quantized in the quantization step with a target code amount in units of the pixel line;
A control step for controlling the quantization step in block units according to the comparison result in the comparison step is executed by a computer,
In the control step, a first quantization step of a block line to which a block to be quantized in the quantization step belongs is determined according to a comparison result in the comparison step, and is quantized in the quantization step. Determining a second quantization step in block units by weighting the first quantization step based on a transform coefficient associated with an upper block of the block;
In the quantization step, the program is quantized by the second quantization step determined in the control step.

Images