JP5873290B2 - Encoder - Google Patents

Description

The present invention relates to a code KaSo location.

  In recent years, digitalization of information related to so-called multimedia such as audio signals and video signals has been advancing rapidly, and in response to this, compression coding and decoding techniques for video signals have attracted attention. The compression encoding / decoding technology is extremely important for the multimedia industry because the storage capacity necessary for storing video signals and the bandwidth required for transmission can be reduced by the compression encoding / decoding technology.

  The compression coding / decoding technique compresses the amount of information (data amount) by using the high autocorrelation (that is, redundancy) of many video signals. The redundancy of the video signal includes temporal redundancy and two-dimensional spatial redundancy. Temporal redundancy can be reduced using block-by-block motion detection and motion compensation. On the other hand, spatial redundancy can be reduced using discrete cosine transform (DCT).

  H. has become one of the international standards. H.264 / MPEG4-AVC employs a context-adaptive coding method called CAVLC or CABAC, and has higher coding efficiency than conventional coding techniques. CAVLC is an abbreviation for Context-based Adaptive-Variable-Length Coding, and CABAC is an abbreviation for Context-Adaptive Binary Arithmetic Coding.

  H. In H.264 / MPEG4-AVC, the maximum code amount per macroblock (the upper limit value of the code amount) is defined. When encoding with 4: 2: 0 color components and 8 bits, the code amount is 3200 bits. It should not be exceeded. Therefore, it is required to perform encoding control so that the code amount does not exceed 3200 bits at the time of encoding. As a result of encoding, if the amount of code exceeds the upper limit, it is a violation of the standard, and it is necessary to re-encode by setting the prediction method and quantization coefficient again. In this case, since the next macroblock cannot be encoded until the re-encoding process is completed, there is a concern about a delay in processing time in a device that requires real-time processing such as a video camcorder.

  In addition to the re-encoding method, there is also a method of encoding using a prediction mode called I_PCM mode in which pixel values are multiplexed as they are. However, in the case of the I_PCM mode, it is necessary to store the I_PCM data in advance because it is not possible to determine whether the code amount exceeds the upper limit value after entropy coding. In order to prevent the code amount per macroblock from exceeding the upper limit value using such a mechanism, there is a technique for predicting the code amount based on a certain threshold from information before entropy coding and replacing it with I_PCM data. It has been proposed (see Patent Document 1).

JP 2007-166039 A

  However, in the technique described in Patent Document 1, it is necessary to pass macroblock image data to the entropy encoding unit. Since it is impossible to know which macroblock in the picture the code amount exceeds the upper limit without actually encoding, in order to be prepared for the possibility that the upper limit will be exceeded in any macroblock, It is necessary to store image data of the entire picture. Therefore, there is a problem that memory consumption increases.

  Further, when CABAC is used as the entropy encoding method, the processing time becomes longer in proportion to the bit rate, so that it is necessary to hold more image data when the bit rate is high. As a result, the memory consumption further increases.

The present invention, when encoding a plurality of images while suppressing an increase in memory consumption, the code amount of each block is intended to make it possible to reduce the possibility of exceeding the upper limit.

The encoding device according to the present invention is based on an acquisition unit that acquires an encoding target block from a first image, and a second image different from the first image, and a predicted image corresponding to the encoding target block Generating means for generating difference information calculating means for calculating difference information between the encoding target block and the predicted image, and index value calculating means for calculating an index value that is an index of the information amount of the difference information Determining means for determining a quantization coefficient so as not to fall below a lower limit value corresponding to the index value; quantization means for quantizing the difference information using the determined quantization coefficient; and the quantization Encoding means for entropy-encoding the difference information, and the correspondence relationship between the index value and the lower limit value is such that the lower limit value is greater than 0 when the index value is greater than or equal to a predetermined threshold value. Associated with the value When the index value is not equal to or greater than the predetermined threshold, the lower limit value is associated with 0, and the determining means determines a provisional quantization coefficient without considering the lower limit value, When the provisional quantization coefficient falls below the lower limit value, the lower limit value is determined as the final quantization coefficient, and when the provisional quantization coefficient does not fall below the lower limit value, the provisional quantization is performed. The coefficient is determined as a final quantization coefficient, and the determination means determines the predetermined number of times that the lower limit value is determined as the final quantization coefficient for the encoding target block acquired from the first image. Before and after exceeding, the correspondence relationship between the index value and the lower limit value is changed .

According to the present invention, when encoding a plurality of images while suppressing an increase in memory consumption can code amount of each block to reduce the possibility of exceeding the upper limit.

1 is a block diagram illustrating a configuration of an imaging apparatus 100 according to a first embodiment. The flowchart which shows the specific example of the process of the quantization control part 115 regarding the specific corresponding | compatible relationship between the index value of difference information and the lower limit value of a quantization coefficient based on 1st Embodiment. The block diagram which shows the structure of the imaging device 300 which concerns on the modification of 1st Embodiment. Processing of quantization control unit 115, replacement control unit 301, and quantization coefficient replacement unit 302 regarding a specific correspondence relationship between the index value of difference information and the lower limit value of the quantization coefficient according to the modification of the first embodiment Flowchart showing a specific example The flowchart which shows the specific example of the process of the quantization control part 115 which concerns on 2nd Embodiment. A block diagram showing a configuration of an imaging apparatus 600 according to a third embodiment. A block diagram showing a configuration of an imaging apparatus 700 according to a fourth embodiment

  Embodiments of the present invention will be described below with reference to the accompanying drawings. The technical scope of the present invention is determined by the claims, and is not limited by the following individual embodiments. In addition, not all combinations of features described in the embodiments are essential to the present invention.

[First Embodiment]
With reference to FIG. 1, an embodiment in which the encoding apparatus of the present invention is applied to an imaging apparatus will be described. In the imaging apparatus 100 shown in FIG. 1, the imaging unit 101 includes a camera unit such as a lens or a CCD, and stores captured moving image data in the frame memory 102 in the order of input. As a result, the frame memory 102 stores the sequence of input images.

The frame memory 102 outputs the sequentially input image data to the inter prediction unit 103 and the intra prediction unit 104 in the encoding order. The inter prediction unit 103 detects, as a motion vector, the amount of motion between the encoding target image and the reference image in units of blocks (for example, macroblocks) including a predetermined number of pixels that are units of encoding or prediction The motion vector thus transmitted is transmitted to the prediction method selection unit 105. Although a specific method for detecting a motion vector in the present embodiment is not particularly limited, a method using the following formula (1) is given as an example.
C = D + λR (1)

  In Expression (1), C is an evaluation value for determining a motion vector. D is an absolute value sum (hereinafter referred to as SATD) of transform coefficients obtained by performing integer transform on difference information obtained by subtracting a coding target image (coding target block) from a predicted image. Alternatively, D may be other types of difference information such as the sum of absolute values of difference information (hereinafter referred to as SAD) obtained by subtracting the encoding target image from the predicted image. In the present embodiment, SATD is used as the difference information D. R is a code amount of a motion vector, λ is a coefficient, and generally a quantization step is used. In this embodiment, a block composed of a predetermined number of pixels, which is a unit of encoding or prediction, is a macroblock composed of 16 pixels × 16 pixels, for example. The code amount R of the motion vector is calculated based on the difference amount between the motion vector of the encoding target macroblock and the estimated motion vector obtained from the motion vectors of the surrounding macroblocks. The final motion vector is detected as the amount of deviation from the encoding target macroblock with respect to the coordinate at which the evaluation value C is minimum within the search range. The inter prediction unit 103 transmits SATD corresponding to the predicted image based on the detected motion vector to the prediction method selection unit 105.

  The intra prediction unit 104 generates a plurality of intra-predicted images using the peripheral pixels of the encoding target macroblock, and performs block matching with the encoding target macroblock. The intra prediction unit 104 selects an intra prediction mode in which the SATD is minimized, and transmits it to the prediction method selection unit 105 together with the SATD at that time.

  The prediction method selection unit 105 selects a motion vector detected by the inter prediction unit 103 and the intra prediction mode obtained by the intra prediction unit 104, and selects the one with the better coding efficiency and transmits it to the predicted image generation unit. Although the selection criteria for selecting a prediction method with good coding efficiency is not particularly limited, in this embodiment, the prediction method selection unit 105 compares the SATD obtained by the inter prediction unit 103 and the SATD obtained by the intra prediction unit 104. The prediction mode with the smaller SATD is selected. The prediction method selection unit 105 transmits the selected prediction mode SATD to the quantization control unit 115.

  Here, the SATD transmitted to the quantization control unit 115 is a value (index value) that serves as an index of the information amount of the difference information between the encoding target macroblock and the predicted image. Accordingly, the inter prediction unit 103, the intra prediction unit 104, and the prediction method selection unit 105 cooperate to perform index value calculation processing.

  The predicted image generation unit 106 generates a predicted image according to the prediction mode selected by the prediction method selection unit 105. The generation of the prediction image is performed based on at least one input image included in the sequence of input images, but the input image that is specifically used differs depending on the prediction mode. For example, when the intra prediction mode is used, the prediction image is generated based only on the input image including the encoding target macroblock. The predicted image generation unit 106 transmits the generated predicted image to the subtracter 107.

  The subtractor 107 calculates difference information (difference image data) between the encoding target macroblock acquired from the frame memory and the predicted image acquired from the predicted image generation unit 106 (that is, the subtracter 107 calculates difference information). Process). The subtractor 107 transmits the calculated difference information to the integer conversion unit 108.

  The integer conversion unit 108 performs integer conversion on the input difference information and transmits the conversion coefficient to the quantization unit 109. The quantization unit 109 quantizes the transform coefficient from the integer transform unit 108 using the quantization coefficient determined by the quantization control unit 115 described later. Since the transform coefficient is derived from the difference information, in a broad sense, the quantization unit 109 quantizes the difference information. The quantization unit 109 transmits the quantized difference information to the entropy encoding unit 114 and the inverse quantization unit 110.

  The inverse quantization unit 110 inversely quantizes the quantized transform coefficient (difference information) and transmits it to the inverse integer transform unit 111. The inverse integer transform unit 111 generates difference information that has not been subjected to integer transform by performing an inverse integer transform process on the inversely quantized transform coefficient. The inverse integer conversion unit 111 transmits the generated difference information to the adder 112.

  The adder 112 adds the difference information generated by the inverse integer transform unit 111 and the predicted image generated by the predicted image generation unit 106. The data after the addition becomes decoded reconstructed image data, which is input to the predicted image generation unit 106 described above and used for generating an intra predicted image. The reconstructed image data is also transmitted to the in-loop filter 113. The in-loop filter 113 performs coding distortion reduction processing on the reconstructed image data, and stores it in the frame memory 102 as reference image data used in inter coding.

  The entropy encoding unit 114 entropy encodes the difference information quantized by the quantization unit 109 (in this embodiment, strictly, the difference information quantized by integer conversion). The entropy encoding unit 114 selects one of a plurality of encoding methods such as CABAC and CAVLC as an encoding method for entropy encoding. The entropy encoding unit 114 outputs code data obtained by entropy encoding as a stream. In addition, the entropy encoding unit 114 notifies the quantization control unit 115 of the generated code amount.

  The quantization control unit 115 determines an appropriate quantization coefficient while monitoring the generated code amount from the entropy coding unit 114 so that the output stream approaches a recording rate set by a CPU (not shown). Notification to the conversion unit 109.

  In the present embodiment, the quantization control unit 115 not only considers the recording rate, but also reduces the possibility that the generated code amount of the entropy encoding unit 114 exceeds the determined maximum code amount (upper limit value). Determine the quantization factor. As a basic idea, the greater the SATD (index value of the information amount of difference information) input from the prediction method selection unit 105, the more the final generated code amount tends to increase. At this time, the smaller the quantization coefficient used by the quantizing unit 109, the smaller the amount of information reduction in the quantizing unit 109, and thus the more likely the final generated code amount exceeds the upper limit. Therefore, the quantization control unit 115 determines the quantization coefficient so as not to fall below the lower limit value corresponding to the index value. Thereby, since the quantization coefficient becomes equal to or higher than the lower limit value, it is expected that the amount of information to be encoded is reduced at least to the extent corresponding to the lower limit value of the quantization coefficient in the quantization unit 109. The possibility that the generated code amount exceeds the upper limit value is reduced.

  As the correspondence relationship between the index value and the lower limit value, an arbitrary correspondence relationship can be adopted depending on the degree of suppressing the possibility that the final generated code amount exceeds the upper limit value, the desired image quality, and the like. For example, a relatively large lower limit value is uniformly associated with an index value that is greater than or equal to a relatively small threshold value, and 0 is uniformly associated with an index value that is less than this threshold value, and the lower limit value is not substantially considered. It is possible. In this case, it is expected that the possibility that the generated code amount exceeds the upper limit value is relatively strongly reduced, but the image quality may be greatly lowered. Alternatively, a relatively small lower limit value is uniformly associated with an index value greater than or equal to a relatively large threshold value, and 0 is uniformly associated with an index value less than this threshold value so that the lower limit value is not substantially considered. It is possible. In this case, although the deterioration of the image quality is suppressed, the possibility that the generated code amount exceeds the upper limit value may not be reduced so much. Alternatively, it is conceivable that a large lower limit value is associated stepwise as the index value increases. In this case, the processing is complicated, but it is expected that suppression of deterioration in image quality and reduction in the possibility that the generated code amount exceeds the upper limit value are achieved in an appropriate balance.

  In this way, there is some correspondence between the index value and the lower limit, and any correspondence can be used as long as the possibility that the generated code amount exceeds the upper limit is partially reduced. Included in the range. Hereinafter, with reference to FIG. 2, a specific example of processing of the quantization control unit 115 regarding a specific correspondence between the index value of the difference information and the lower limit value of the quantization coefficient will be described.

  First, in S200, the quantization control unit 115 determines the quantization coefficient Qp based on the set recording rate and the generated code amount of the entropy encoding unit 114. At this point, Qp is a provisional quantization coefficient, and the final quantization coefficient is determined in S203 or S204 described later.

  In S201, the quantization control unit 115 determines whether the SATD input from the prediction method selection unit 105 is equal to or greater than a predetermined threshold Th0. If it is equal to or greater than the threshold value (SATD ≧ Th0), the process proceeds to S202, and if not (SATD <Th0), the process proceeds to S204.

  In S202, the quantization control unit 115 determines whether or not Qp determined in S200 is lower than the lower limit value Qp0. If lower (Qp <Qp0), the process proceeds to S203, and if not (Qp ≧ Qp0), the process proceeds to S204.

  As described above, the lower limit value of the quantization coefficient is a value corresponding to SATD (index value). However, in the example of FIG. 2, when SATD ≧ Th0, the lower limit value is uniformly Qp0, and SATD <Th0. In this case, the lower limit value is not considered (for example, set to 0).

  In S203, the quantization control unit 115 replaces Qp with Qp0, and notifies the quantization unit 109 of Qp0. On the other hand, in S204, the quantization control unit 115 notifies the quantization unit 109 of Qp as it is. Thus, the final quantization coefficient is determined so as not to fall below the lower limit value Qp0.

  As described above, according to the present embodiment, the quantization control unit 115 does not lower the lower limit corresponding to the index value (for example, SATD) of the information amount of the difference information to be encoded. Is notified to the quantization unit 109. The configuration of the present embodiment can be implemented by adding a few circuits such as a comparator, and does not require an extra large memory capacity. Therefore, according to the present embodiment, when the sequence of the input image is encoded in units of blocks including a predetermined number of pixels, the code amount of each block can exceed the upper limit while suppressing an increase in memory consumption. Can be reduced.

  In the present embodiment, SATD is used as an example of the index value, but SAD, variance, or the like may be used instead. That is, any kind of information may be used as long as it is an index of the information amount of the difference information to be encoded.

  Further, as described above, the correspondence relationship between the index value and the lower limit value is not limited to the above. As another example, a correspondence relationship including three types of threshold values related to SATD and three types of corresponding lower limit values will be described. The threshold values related to SATD are SATD_Th0, SATD_Th1, and SATD_Th2 (SATD_Th0> SATD_Th1> SATD_Th2), and the lower limit values are Qp0, Qp1, Qp2 (Qp0> Qp1> Qp2). SATD_Th0 corresponds to Qp0, SATD_Th1 corresponds to Qp1, and SATD_Th2 corresponds to Qp2. Therefore, when SATD> SATD_Th2 and Qp <Qp2, Qp is replaced with Qp2. Similarly, when SATD> SATD_Th1 and Qp <Qp1, Qp is replaced with Qp1, and when SATD> SATD_Th0 and Qp <Qp0, Qp is replaced with Qp0. In this way, by changing the lower limit value stepwise according to the index value, it is possible to prevent the quantization coefficient from becoming too large and to suppress deterioration in image quality.

  Further, the correspondence relationship between the index value and the lower limit value may be changed according to the type of the encoding target macroblock. For example, depending on whether or not the encoding target macroblock is an intra macroblock (that is, whether or not the prediction image is generated based on only the input image to which the encoding target macroblock belongs in the prediction image generation unit 106). Can change the correspondence. How the correspondence is changed is arbitrary, and may be changed as appropriate in consideration of the desired image quality.

  Similarly, the correspondence between the index value and the lower limit value may be changed according to the encoding method (for example, CABAC or CAVLC) selected by the entropy encoding unit 114. Thereby, the difference in correlation between the index value (for example, SATD) according to the encoding method and the generated code amount can be taken into consideration, and the lower limit value of the quantization coefficient can be set more accurately. As a result, it is possible to suppress degradation of image quality and to reduce the possibility that the generated code amount exceeds the upper limit value. Also in this case, how to change the correspondence is arbitrary, and may be changed as appropriate in consideration of the desired image quality.

(Modification)
With reference to FIG. 3 and FIG. 4, another configuration in which the quantization coefficient can be determined so as not to fall below the lower limit value corresponding to the information amount index value (for example, SATD) of the difference information to be encoded will be described. To do.

  FIG. 3 is a block diagram illustrating a configuration of an imaging apparatus 300 according to a modification example of the first embodiment. 3, the same or similar components as those in FIG. 1 are denoted by the same reference numerals, and the description thereof is omitted.

  The imaging device 300 includes a replacement control unit 301 and a quantization coefficient replacement unit 302 in addition to the configuration of the imaging device 100. In addition, the prediction method selection unit 105 transmits SATD to the replacement control unit 301 instead of the quantization control unit 115.

  Next, referring to FIG. 4, specific processing of the quantization control unit 115, the replacement control unit 301, and the quantization coefficient replacement unit 302 regarding a specific correspondence between the index value of the difference information and the lower limit value of the quantization coefficient An example will be described. In FIG. 4, steps in which the same or similar processes as those in FIG.

  In FIG. 4, S401 to S403 are added between S201 and S202 and S204. Also, unlike the case of FIG. 2, the processing of S201 is executed by the replacement control unit 301, and the processing of S202 to S204 is executed by the quantization coefficient replacement unit 302.

  When it is determined in S201 that SATD ≧ Th0, in S401, the replacement control unit 301 sets the replacement flag held by the quantization coefficient replacement unit 302 to on. On the other hand, if it is determined in S201 that SATD <Th0, the replacement control unit 301 sets the replacement flag held by the quantized coefficient replacement unit 302 to off in S402.

  In S403, the quantization coefficient replacement unit 302 determines whether the replacement flag is set to on. If it is set to on, the process proceeds to S202; otherwise, the process proceeds to S204. Thereby, the lower limit value Qp0 of the quantization coefficient is considered only when SATD ≧ Th0, and Qp determined in S200 is replaced with Qp0 as necessary. Therefore, as in FIG. 2, the final quantization coefficient is determined so as not to fall below the lower limit corresponding to SATD.

[Second Embodiment]
In the second embodiment, before and after the number of times that the quantization coefficient replacement has occurred for a predetermined picture (input image) (that is, the lower limit value is determined as the final quantization coefficient) exceeds the predetermined number, the index The correspondence between the value and the lower limit value changes. In the present embodiment, the configuration of the imaging apparatus 100 is the same as that of the first embodiment (see FIG. 1), but the operation of the quantization control unit 115 is different from that of the first embodiment.

  FIG. 5 is a flowchart illustrating a specific example of processing of the quantization control unit 115 according to the second embodiment. In FIG. 5, steps in which the same or similar processing as in FIG.

  In S501, the quantization control unit 115 determines whether or not the variable MBNum is less than the threshold value Th. MBNum is initialized to 0 each time the encoding target picture is switched, and indicates the number of times Qp replacement has occurred for a specific encoding target picture. If MBNum is less than Th, the process proceeds to S201; otherwise, the process proceeds to S503.

  The process of S502 is the same as that of S203 of FIG. 2, but the quantization control unit 115 adds 1 to MBNum in addition to replacing Qp with Qp0. As a result, the number of times Qp replacement has occurred is counted.

  For the encoding target macroblock to be encoded after MBNum is equal to or greater than Th, the processing of S503 to S505 and S204 is executed. The processing from S503 to S505 is the same as S201, S202, and S502, except that Th1 is used instead of Th0, and Qp1 is used instead of Qp0. Accordingly, the correspondence between the index value (in this example, SATD) and the lower limit value changes.

  The large number of occurrences of Qp substitution means that there is a high possibility that the SATD is large and the amount of generated codes is large. Therefore, according to the processing of the present embodiment, it is possible to determine whether or not a picture includes many macroblocks having a large SATD. When it is found that the picture includes many macroblocks, the correspondence between the SATD and the lower limit value is changed. By doing so, the final determination method of Qp can be changed. The specific correspondence to be used can be appropriately determined according to the desired image quality. Therefore, for example, Th0> Th1 may be satisfied, or Th0 <Th1 may be satisfied. Alternatively, for example, Th0 = Th1, and Qp0 and Qp1 may be different.

[Third Embodiment]
In the third embodiment, a predicted value of a generated code amount corresponding to an index value (for example, SATD) of difference information to be encoded is used. FIG. 6 is a block diagram illustrating a configuration of an imaging apparatus 600 according to the third embodiment. In FIG. 6, the same or similar components as those of the modification of the first embodiment (see FIG. 3) are denoted by the same reference numerals, and the description thereof is omitted.

  The imaging apparatus 600 includes a threshold value changing unit 601 in addition to the configuration of the imaging apparatus 300. In addition to the replacement control unit 301, the prediction method selection unit 105 transmits SATD to the threshold value changing unit 601. Furthermore, the entropy encoding unit 114 notifies the threshold change unit 601 of the generated code amount (actual generated code amount) in addition to the quantization control unit 115. In the present embodiment, this generated code amount is referred to as “CODE”.

  The threshold value changing unit 601 holds in advance a table that stores the predicted value of the generated code amount for the encoding target macroblock corresponding to each of the plurality of SATDs. In the present embodiment, the prediction value corresponding to the SATD received from the prediction method selection unit 105 is referred to as “CODE0”.

  The threshold value changing unit 601 compares CODE0 and CODE. When CODE is larger than CODE0, there is a high possibility that the actual generated code amount (that is, CODE) exceeds the upper limit. Therefore, the threshold value changing unit 601 instructs the replacement control unit 301 to decrease the threshold value Th0 used by the replacement control unit 301 in S201 of FIG. As a result, substitution of Qp is likely to occur, and the possibility that CODE exceeds the upper limit value is reduced.

  On the other hand, when CODE is smaller than CODE0, there is a high possibility that there is a margin between the actual generated code amount (that is, CODE) and the upper limit value. Therefore, the threshold value changing unit 601 instructs the replacement control unit 301 to increase the threshold value Th0. As a result, Qp substitution hardly occurs and the image quality is improved.

  When CODE is approximately the same as CODE0, the threshold value changing unit 601 does not change the threshold value Th0.

  By the above processing, for example, if CODE = 250 when CODE0 is 200 when SATD = 1000, Th0 is small, and if CODE0 is 200 when SATD = 1000, Th is large. Become. Also, if CODE0 is 200 when SATD = 1000, and CODE = 205, Th does not change.

  Note that the change of the threshold Th0 described here is only an example. Changing the threshold Th0 substantially means changing the correspondence between the index value and the lower limit value. Therefore, when this embodiment is generalized, when CODE> CODE0, the correspondence relationship between the index value and the lower limit value changes so that the smaller index value corresponds to the larger lower limit value. When CODE <CODE0, the correspondence relationship between the index value and the lower limit value changes so that the larger index value corresponds to the smaller lower limit value. Thereby, the possibility that the generated code amount exceeds the upper limit is reduced, and the image quality is improved.

[Fourth Embodiment]
In the fourth embodiment, instead of setting the lower limit value of the quantization coefficient, by reducing at least one of the DC component (DC component) and AC component (AC component) of the difference information to be encoded, The possibility that the generated code amount exceeds the upper limit is reduced.

  FIG. 7 is a block diagram illustrating a configuration of an imaging apparatus 700 according to the fourth embodiment. In FIG. 7, the same or similar components as those in the first embodiment (see FIG. 1) are denoted by the same reference numerals, and description thereof is omitted.

  The imaging apparatus 700 includes a maximum code amount excess prevention unit 701 and a DC / AC component truncation unit 702 in addition to the configuration of the imaging apparatus 100. Also, the prediction method selection unit 105 transmits an index value (for example, SATD) to the maximum code amount excess prevention unit 701 instead of the quantization control unit 115. Further, the quantization control unit 115 notifies the quantization coefficient to the maximum code amount excess prevention unit 701 in addition to the quantization unit 109.

  The maximum code amount excess prevention unit 701 maintains a table that stores a SATD threshold (SATD0) corresponding to each of a plurality of quantization coefficients. Then, the maximum code amount excess prevention unit 701 identifies SATD0 corresponding to the quantization coefficient notified from the quantization control unit 115, and performs the following comparison regarding the SATD received from the prediction method selection unit 105. Maximum code amount excess prevention unit 701 notifies DC / AC component truncation unit 702 of the flag determined as follows based on the comparison result.

(1) When SATD0> SATD, no flag is set (2) When SATD0 ≦ SATD <SATD0 + α, flag 1 is set (3) When SATD0 + α ≦ SATD <SATD0 + β, flag 2 is set (4) When SATD0 + β ≦ SATD <SATD0 + γ Set flag 3 (5) When SATD0 + γ ≦ SATD Set flag 4 Note that α, β, and γ are natural numbers, and α <β <γ.

  The DC / AC component truncation unit 702 receives the difference information (transform coefficient) frequency-converted and quantized from the quantization unit 109, and based on the flag notified from the maximum code amount excess prevention unit 701, as follows. Reduce the amount of information.

(1) When the flag is not set No reduction is performed (2) When the flag 1 is set Cut a part of the AC component of the color difference (3) When the flag 2 is set The AC component and luminance of the color difference Cut a part of AC component (4) When flag 3 is set AC part and DC component of color difference and part of AC component of luminance are cut (5) When flag 4 is set AC of color difference Cut a part of the component and DC component, and the AC component and DC component of luminance.

  When the AC / DC component is cut, the possibility that the generated code amount exceeds the upper limit value decreases, but the image quality deteriorates. Therefore, in this embodiment, priority is given to the AC / DC component to be cut as described above, and image quality degradation is minimized by gradually cutting from the component that has little influence on the image quality according to SATD. The possibility that the generated code amount exceeds the upper limit value is reduced while limiting to the limit. That is, when the SATD is relatively small as in the case where the flag 1 is set, only the AC component of the color difference is cut. When the SATD is relatively large as in the case where the flag 4 is set, not only the color difference AC component but also the color difference DC component, the luminance AC component, and the luminance DC component are cut.

  Note that the information amount reduction method described above is merely an example. When generalized, in this embodiment, the DC / AC component truncation unit 702 reduces the information amount of at least one of the direct current component and the alternating current component of the frequency-converted difference information (conversion coefficient). At this time, the DC / AC component truncation unit 702 reduces the amount of information as the index value (eg, SATD) increases. As a result, it is possible to reduce the possibility that the code amount of each macroblock exceeds the upper limit value while suppressing the deterioration of the image quality.

  As in the first embodiment, not only SATD but also SAD or variance can be used as an index value.

[Other Embodiments]
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

Claims (3)

Obtaining means for obtaining a block to be encoded from the first image;
Generating means for generating a predicted image corresponding to the encoding target block based on a second image different from the first image;
Difference information calculating means for calculating difference information between the encoding target block and the predicted image;
Index value calculating means for calculating an index value that is an index of the information amount of the difference information;
Determining means for determining a quantization coefficient so as not to fall below a lower limit corresponding to the index value;
Quantization means for quantizing the difference information using the determined quantization coefficient;
Encoding means for entropy encoding the quantized difference information ;
Have
The correspondence relationship between the index value and the lower limit value is that the lower limit value is associated with a value greater than 0 when the index value is greater than or equal to a predetermined threshold, and the index value is not greater than or equal to the predetermined threshold. Is a relationship in which the lower limit value is associated with 0,
The determining means determines a provisional quantization coefficient without considering the lower limit value, and determines the lower limit value as a final quantization coefficient when the provisional quantization coefficient falls below the lower limit value. And determining the provisional quantization coefficient as a final quantization coefficient when the provisional quantization coefficient does not fall below the lower limit value,
The determination means, for the encoding target block acquired from the first image, before and after the number of times when the lower limit value is determined as a final quantization coefficient exceeds a predetermined number of times, the index value and the lower limit value, An encoding device characterized by changing a correspondence relationship between the two . The determining unit changes a correspondence relationship between the index value and the lower limit value according to whether or not the generating unit has generated the predicted image based only on the first image. The encoding device according to claim 1. The encoding means entropy encodes the quantized difference information by an encoding method selected from a plurality of encoding methods including CABAC and CAVLC,
It said determining means, the encoding apparatus according to claim 1 or 2, characterized in that changing the relationship between the lower limit value and the index value according to the selected coding scheme.

Images