JP2005236710A - Moving image encoder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a moving image encoder for improving encoding efficiency and image quality by selecting an M value corresponding to the characteristics of a desired image and reducing a processing cost even under the condition of a low bit rate. <P>SOLUTION: In a motion vector information detecting unit 114, the largeness of a motion vector sought by a motion compensation predicting unit 113 and a retrieving range used when the motion vector is sought are made correlated with the largeness of the motion vector of each M value and the retrieving range, and then an optimum M value is determined by using generation frequency of motion vectors located in the retrieving range. In an optimum M value determining unit 116, a fluctuation of the optimum M value is controlled by using a ratio of the generated code amount of the motion vector sought by the motion compensation predicting unit 113, to the whole generated code amount of a current encoding picture obtained by a generated code amount counter 115 and held by a buffer unit 107; and finally an optimum M value for next encoding is determined. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a moving image encoding apparatus, and more particularly to a moving image encoding apparatus that encodes moving image information while performing code amount control by an inter-frame predictive encoding method.

  Examples of compression coding methods for image information include internationally standardized H.261, H.262, H.263, H.264, MPEG (Moving Picture Experts Group) -1, MPEG-2, MPEG-4, and the like. Are known. These compression coding methods include processes such as DCT (Discrete Cosine Transform) and motion compensation interframe prediction, and more efficiently increase the compression rate, improve the image quality, especially at low bit rates. There is a demand for improving image quality.

  As a compression encoding method for image information, there is an inter-frame prediction encoding method as one of the methods using compression in the time axis direction. In MPEG-1, MPEG-2, MPEG-4, etc. Both forward inter-picture prediction coding (hereinafter referred to as forward prediction) and backward inter-picture prediction coding (hereinafter referred to as backward prediction) can be used. Forward prediction is to detect the motion between the past frame and the current frame and generate the current frame from the past frame, and backward prediction is the future frame and the current frame. Is detected, and the current frame is generated from the future frame.

  An image encoded by the forward prediction that can be a reference image is a P picture, and an image encoded by the bidirectional prediction that is a referenced image using both the forward prediction and the backward prediction is a B picture. An image of intra-frame coding (intra-picture coding) is an I picture. In the MPEG-1, MPEG-2, MPEG-4, etc., the above-mentioned pictures are encoded by combining intra-frame encoding and prediction encoding in the order of IBBPBBP. .

  Here, an interval between adjacent I and P pictures or between adjacent P pictures is generally called an M value (in this specification, the M value means this picture interval). For example, when M = 3, there are two B pictures between adjacent I and P pictures or between adjacent P pictures, such as IBBPBBP. In general, when encoding using MPEG technology, the M value is often fixed.

  However, in the case of a fast-moving image, if the distance between pictures increases, it is necessary to increase the motion vector search range in order to accurately capture the motion, which increases processing costs. On the other hand, if the search range is reduced in order to reduce the processing cost, the accuracy of the motion vector is lowered, the prediction efficiency is lowered, and the coding efficiency is also lowered. Therefore, in order to reduce the processing cost and prevent a decrease in encoding efficiency, when encoding an image using MPEG technology, the M value is not fixed and is encoded according to the characteristics of the motion. 2. Description of the Related Art Conventionally, a moving image coding apparatus that performs coding by changing the M value is known (see, for example, Patent Documents 1 and 2).

  For example, Patent Literature 1 discloses a method of increasing the inter-P-picture distance according to the motion vector distribution ratio and reducing the inter-frame distance by the inter-picture prediction difference value. That is, as shown in FIG. 6, when considering a motion vector search range of X13 to Xu3 and encoding the current picture with M = 2, X12 to Xu2 with the search range set to 2/3 The next M value is set to 3 if the ratio of the motion vector distribution detected in the range is large, and if the ratio is small, the M value is maintained at 2 and the prediction error between pictures exceeds a certain value. In this case, the encoding efficiency is improved by decreasing the M value from 2 to 1, determining the M value to be encoded next, and making the M value variable. According to the conventional moving picture coding apparatus described in Patent Document 1, the coding efficiency can be improved by changing the M value according to the feature of the moving picture as compared with the case where the M value is fixed.

  Patent Document 2 discloses a method for increasing / decreasing the interframe distance from the average value of the magnitude of the motion vector and the magnitude of the temporal change of the motion vector. According to the conventional moving picture encoding device described in Patent Document 2, the M value is made variable as in the conventional device described in Patent Document 1, thereby improving the encoding efficiency as compared with the case where the M value is fixed. be able to.

JP-A-9-294266 JP 2000-333179 A

  The conventional moving image coding apparatus described in Patent Document 1 determines the optimum M value by using the magnitude of the motion vector obtained at the time of prediction, and increases the M value when the magnitude of the motion vector is small. To do. In addition, the conventional moving image encoding device described in Patent Document 2 determines increase / decrease in the M value using the average of the magnitudes of the motion vectors and the temporal change of the motion vector, and the conventional moving image described in Patent Document 1 is used. Similar to the encoding device, when the size of the motion vector is small, the average value of the motion vector is small, so the M value is large. When the average value is large, the M value is small.

  For example, when considering an image of a water surface that is gently shaking, this image has the feature that the change in motion is small and complicated. Also, as a feature of encoding, when the motion vector required for predicting the above image is small and the motion is complicated, the motion vector varies, and the motion vector is encoded. The amount of codes may increase. When the code amount of the motion vector increases, under the low bit rate condition, the code amount given in addition to the motion vector may be insufficient, resulting in a decrease in encoding efficiency and image quality.

  However, the conventional moving picture coding apparatuses described in Patent Documents 1 and 2 both require a small motion vector size when coding an image such as the gently fluctuating water surface. Although the M value is increased, when the bit rate is low, increasing the M value increases the generated code amount of the motion vector, resulting in a decrease in encoding efficiency and a deterioration in image quality.

  The present invention has been made in view of the above points. The M value is selected according to the characteristics of an arbitrary image, and even under a low bit rate condition, the processing cost is further reduced and the encoding efficiency is improved and the image quality is improved. It is an object of the present invention to provide a moving image encoding apparatus that can be used.

  In order to achieve the above object, the first invention is obtained by encoding a moving image signal by three kinds of encoding methods of intra-picture coding, forward inter-picture predictive coding, and bi-directional inter-picture predictive coding. An encoded signal obtained by combining an I picture, a P picture, and a B picture, or an I picture and P obtained by encoding a moving picture signal by two encoding methods of intra-picture encoding and previous inter-picture predictive encoding A video encoding apparatus based on an inter-picture predictive encoding system that outputs an encoded signal combined with a picture, and an interval between adjacent I and P pictures in the encoded signal, or two adjacent P A moving picture coding apparatus that performs coding by changing an M value indicating a picture interval and performs motion compensation based on reference picture data obtained by local decoding of a coded signal and a moving picture signal To find the motion vector Compensation code predicting means, generated code quantity counting means for determining the ratio of the generated code quantity of the motion vector obtained by the motion compensation predicting means to the total generated code quantity per one encoded current picture, and motion compensated prediction means When the magnitude of the motion vector, the search range of the motion vector, and the predicted absolute value error are received as input, and the currently encoded picture is a P picture, the magnitude of the input current M-value motion vector The search range and the number of motion vectors existing in the search range are determined for each of a plurality of different M values determined in advance with reference to the search range used when the motion vector is obtained, and the motion vector The motion vector information detecting means for selecting the M value corresponding to the search range having the largest number of occurrences as the optimum M value and the generated code amount counting means. When the ratio exceeds the threshold, the final optimum M value when the M value selected by the motion vector information detecting means is further reduced is encoded next. And when the ratio is equal to or less than the threshold value, the optimum M value determining means for determining the final optimum M value when the M value selected by the motion vector information detecting means is encoded next is included. Based on a typical optimum M value, a picture to be encoded next is designated and encoded.

  In the present invention, when the currently encoded picture is a P picture, the number of motion vectors generated within a predetermined search range for each of a plurality of different M values is obtained, and the number of motion vectors generated is calculated. The M value corresponding to the most search range is selected as the optimal M value, and the ratio of the generated code amount of the motion vector to the total generated code amount per encoded current picture is compared with a predetermined threshold, When the ratio exceeds the threshold value, the M value obtained by further reducing the selected M value is determined as the final optimum M value for the next encoding. The determined M value is determined as the final optimum M value when the next encoding is performed. Therefore, by using the motion vector search range, it is possible to select an M value corresponding to the feature of an arbitrary image. Both , By considering the percentage of the generated code amount of motion vectors in the total generated code amount per one current picture, it is possible to obtain an accurate M values even at low bit rates.

  In order to achieve the above object, according to a second aspect of the present invention, the motion vector information detecting means according to the first aspect of the present invention encodes a B picture when the currently encoded picture is a B picture. It is determined whether or not there are a predetermined number or more of the forward motion vectors obtained in the search range defined by the optimum M value selected when the P picture is encoded immediately before, and the B picture is encoded. When there are a predetermined number or more of the motion vectors in the forward direction, the optimum M value selected when the P picture is encoded immediately before is selected as the optimum M value as it is. It has a configuration further comprising means for selecting the M value calculated in the B picture instead of the optimum M value as the optimum M value for the next encoding.

  In the present invention, when there are no more than a predetermined number of forward motion vectors of the currently encoded B picture within the search range determined by the optimal M value determined when the P picture is encoded, the optimal M value Since the M value calculated for the B picture is used as the M value of the next picture to be encoded, the optimal M value can be corrected using the motion vector information of the B picture.

  According to the present invention, the motion vector search range and the size of the motion vector are made to correspond to the search range of each M value and the size of the motion vector, and the optimum M value is determined according to the occurrence frequency of the motion vector existing in the search range. By controlling, it is possible to select an M value according to the characteristics of an arbitrary image, so that it is possible to reduce processing costs and prevent a decrease in encoding efficiency.

  Further, according to the present invention, since the increase / decrease in the optimum M value is controlled in accordance with the ratio of the generated code amount of the motion vector to the total generated code amount per current picture, it is accurate even at a low bit rate. The M value can be obtained, and further reduction in coding efficiency can be prevented at a low bit rate.

  Furthermore, according to the present invention, when there are no more than a predetermined number of motion vectors in the forward direction of the currently encoded B picture within the search range determined by the optimum M value determined when the P picture is encoded. Since the M value calculated in the B picture instead of the optimal M value is used as the M value of the next picture to be encoded, the optimal M value is corrected using the motion vector information of the B picture. Compared with the case where the next optimum M value is determined only by the picture, the optimum M value can be selected with higher accuracy.

  Next, the best mode for carrying out the present invention will be described with reference to the drawings. In the present invention, when an M value suitable for an arbitrary moving image is obtained, the M value is obtained without performing motion detection calculation as preprocessing and using data obtained at the time of encoding. It is possible to obtain high image quality even under a low bit rate condition with little computation.

  FIG. 1 is a block diagram showing an embodiment of a moving picture encoding apparatus according to the present invention. The moving image coding apparatus according to the present embodiment includes an image input unit 101, an image memory unit 102, a subtractor 103, a DCT unit 104, a quantization unit 105, a variable length coding (VLC) unit 106, a buffer unit 107, a code Quantity control unit 108, inverse quantization unit 109, inverse DCT unit 110, adder 111, motion compensation prediction unit 113, motion vector information detection unit 114, generated code amount count unit 115, optimum M value determination unit 116, M value control Part 117 and a bit stream output part 118.

  The image input unit 101 inputs moving image data to be encoded one by one, and outputs the required number of moving image data to the image memory unit 102. The image memory unit 102 can hold several pieces of moving image data sent from the image input unit 101, and outputs moving image data to be encoded designated by the M-value control unit 117.

  The subtracter 103 subtracts the motion compensation signal output from the motion compensation prediction unit 113 from the moving image data output from the image memory unit 102, and outputs a difference signal. However, only when the moving image data output from the image memory unit 102 is an I picture, the subtractor 103 passes the moving image data and inputs the moving image data to the DCT unit 104. In the DCT unit 104, in order to obtain higher coding efficiency, the above-described differential signal or moving image data to be taken as an I picture is converted into discrete cosine transform (DCT) as an example of orthogonal transform means in units of about 8 × 8 pixels. To convert to a coefficient component.

  The quantization unit 105 receives the signal (coefficient component) converted by the DCT unit 104 as an input, and quantizes each coefficient component in a quantization step controlled by the code amount control unit 108. The VLC unit 106 receives the signal quantized by the quantization unit 105, converts the signal into a variable length code such as a Huffman code, and outputs the variable length digital data to the buffer unit 107. The buffer unit 107 holds data that has been variable-length encoded by the VLC unit 106 and data that has been generated during encoding. The data held in the buffer unit 107 is output as encoded image data from the bit stream output unit 118. The encoded image data is an encoded signal sequence that combines an I picture, a P picture, and a B picture, such as IBBPBBP, but is an encoded signal sequence that combines an I picture and a P picture. May be.

  The code amount control unit 108 controls the quantization step of the quantization unit 105 in order to control the code amount for encoding the next picture based on the data held in the VLC unit 106. . The inverse quantization unit 109 performs a conversion process opposite to that of the quantization unit 105 for the prediction process on the quantized signal output from the quantization unit 105. The inverse DCT unit 110 performs inverse orthogonal transform on the signal inversely quantized by the inverse quantization unit 109 and inversely to the DCT unit 104.

  The adder 111 adds the prediction error signal obtained by the motion compensation prediction unit 113 to the signal output from the inverse DCT unit 110 to generate a reference image signal that is encoded image data. However, the adder 111 does not perform the addition operation on the image data to be the I picture, and passes the decoded image data from the inverse DCT unit 110 and supplies it to the decoded image memory unit 112.

  The decoded image memory unit 112 holds the decoded image data supplied from the adder 111. The motion compensation prediction unit 113 compares the moving image data to be encoded supplied from the image memory unit 102 with the reference image data obtained by local decoding supplied from the decoded image memory unit 112, and performs motion compensation. To obtain a motion vector.

  The motion vector information detection unit 114 uses the search range and the predicted absolute value error used when the motion vector size and motion vector obtained by the motion compensation prediction unit 113 are obtained, as described in detail later. The optimum M value is determined based on the magnitude of the motion vector and the occurrence frequency of the motion vector existing in the search range, and the result of the optimum M value is output to the optimum M value determining unit 116. The generated code amount counting unit 115 is based on the total generated code amount of the encoded current encoded picture held in the buffer unit 107 and the generated code amount of the motion vector obtained by the motion compensation prediction unit 113. The ratio (occupation ratio) of the code amount of the motion vector in the area is obtained.

  The optimum M value determination unit 116 occupies the ratio of the generated code amount of the motion vector obtained by the motion compensation prediction unit 113 to the total generated code amount of the current encoded picture obtained by the generated code amount counting unit 115. By controlling the increase / decrease of the optimum M value as will be described in detail later, the optimum M value to be encoded next is finally determined. The M value control unit 117 designates the picture type and picture to be encoded next to the image memory unit 102 from the optimal M value determined by the optimal M value determination unit 116, and I pictures at regular intervals. specify.

  Next, the operation of the motion vector information detection unit 114 will be described in detail together with a flowchart for obtaining the number of motion vectors within a search range having a motion vector when the M value of the present invention shown in FIG. 2 is made variable. . First, the motion vector information detection unit 114 receives the horizontal and vertical sizes of the motion vector obtained by the motion compensation prediction unit 113, the size of the search range, the size of the predicted absolute value error, and the picture type. The process for determining the optimum M value is started (step S201). Subsequently, the motion vector information detection unit 114 determines whether the currently encoded picture is a P picture or a B picture based on the picture type input in step S201 (step S202).

  When the motion vector information detection unit 114 determines in step S202 that the current encoded picture is a P picture, the motion vector information detection unit 114 determines the current M value based on the size of the search range of the current M value input in step S201. Based on the value, a search range corresponding to the distance of each M value is obtained (step S203), and each M value in step S203 is determined based on the magnitude of the horizontal and vertical motion vectors input in step S201. Similarly to the case where the search range corresponding to is obtained, the magnitude of the motion vector corresponding to each M value is obtained (step S204).

  Subsequently, the motion vector information detection unit 114 uses the search range corresponding to each M value obtained in steps S203 and S204 and the magnitude of the motion vector to count the number of motion vectors that exist in each search range. (Step S205). Next, among the number of motion vectors generated in each search range obtained in step S205, the M value corresponding to the search range with the highest occurrence frequency is the interval between P pictures to be encoded next. The optimum M value is selected (step S206).

  Subsequently, when the optimum M value selected in step S206 is equal to the M value of the currently encoded picture, the motion vector information detection unit 114 inputs a predetermined value determined in advance for each M value in step S201. The magnitude of the predicted absolute value error is compared (step S207), and if the magnitude of the predicted absolute value error is greater than or equal to a predetermined value predetermined for each M value, the current M value is If it is determined that the predicted absolute value error is smaller than the predetermined value, the current M value is determined as the optimum M value (step S208).

  On the other hand, if the optimum M value selected in step S206 is different from the M value of the currently encoded picture, the selected optimum M value is determined as the next optimum M value (step S208). The motion vector information detection unit 114 outputs the optimum M value determined in step S208 to the external optimum M value determination unit 116 (step S209), and uses it in step S210.

  If the motion vector information detection unit 114 determines in step S202 that the current encoded picture is a B picture, the motion vector information detection unit 114 holds the optimum M value determined using the P picture in step S208 (step S210). ). Subsequently, as in step S203, the size of the search range corresponding to the optimum M value determined using the P picture held in step S210 is obtained (step S211).

  Further, the size of the forward motion vector of the B picture corresponding to the search range obtained in step S211 is obtained by the same method as in step S204 (step S212). Subsequently, using the search range obtained in step S211 and the forward motion vector of the B picture obtained in step S212, the number of forward motion vectors of the B picture in the search range is counted. (Step S213).

  Next, the ratio of the number of occurrences obtained in step S213 to the number of occurrences obtained in step S206 is obtained. If the ratio is equal to or greater than a certain threshold value (hereinafter referred to as threshold F), it is obtained in step S208. The M value is determined as the optimum M value, and if the ratio is smaller than F, the M value in the B picture different from the M value obtained in step S208 is determined as the optimum M value (step S214).

  That is, if the above ratio is equal to or greater than the threshold value F, it means that there are a large number of forward motion vectors of B pictures in the search range, and therefore the M value corresponding to the search range is appropriate. Therefore, the M value determined in the P picture obtained in step S208 is determined as the optimum M value. On the other hand, the fact that the above ratio is smaller than the threshold value F means that the number of motion vectors in the forward direction of the B picture is small in the search range, and the motion vectors are scattered appropriately in any search range. This means that the M value determined in the P picture obtained in step S208 is determined to be not optimal, and thus is smaller than the M value calculated in the current B picture, for example, the M value determined in the P picture by a predetermined value. The M value is determined as the optimum M value.

  Finally, the motion vector information detection unit 114 outputs the optimal M value determined in step S214 to the external optimal M value determination unit 116 (step S209). As described above, according to the present embodiment, the optimum M value is corrected using the motion vector information of the B picture, so that the optimum M value can be optimized with higher accuracy than in the case where the next M value is determined only by the P picture. The M value can be determined.

  Next, an example of how to obtain the search range and motion vector for each M value in the motion vector information detection unit 114 will be described with reference to FIG. The motion vector information detection unit 114 obtains a search range corresponding to each M value (M = 2, 3, 4) as shown in FIG. 3 with reference to the search range of M = 2. If the search range when M = 2 is a square of one side R indicated by 301 in FIG. 3, the search range when M = 3 is a square of one side R / 2 indicated by 302 in FIG. The search range in this case is obtained as a square of one side R × 2/3 indicated by 303 in FIG.

  The motion vector information detection unit 114 counts the number of motion vectors per currently encoded picture generated within each search range obtained for each M value, and determines the most motion vector number. The M value corresponding to a large search range is determined as the interval between P pictures to be encoded next. For example, assuming that there are 100 motion vectors in total, there are 10 motion vectors in the search range of M = 1, 10 motion vectors in the search range of M = 2, and 60 motion vectors in the search range of M = 3. When there are 20 motion vectors in the search range of M = 4, it is determined that the search range of M = 3 is appropriate, and M = 3 is set as the optimum M value.

  However, when the current M value and the optimum M value are equal, the current M value is reduced by using the magnitude of the prediction error of the current encoded picture, and when it is equal to or greater than a certain value, If it is smaller than the value, the current M value is set as the optimum M value.

  Next, the operation of the optimum M value determination unit 116 of FIG. 1 will be described in detail with reference to the flowchart of FIG. The flowchart of FIG. 4 is activated for each picture. The optimum M value determination unit 116 holds the occupation rate (ratio) of the generated code amount of the motion vector per picture supplied from the generated code amount count unit 115 of FIG. Data on the occupancy rate of the motion vector generated code amount in the generated code amount for the current M value and the number of generated motion vectors for each search range are input (step S401).

  Subsequently, the optimal M value determination unit 116 determines increase / decrease in the M value using the occupation rate of the motion vector generation code amount input in Step S401, and determines whether or not to decrease the M value (Step S401). S402). That is, in step S402, when the occupation rate of the generated code amount of the motion vector exceeds a predetermined threshold value compared with the code amount per picture, the current M value (P picture interval) Since the code amount given to the motion vector may be small and the code amount given to other than the motion vector may be reduced, it is considered that the image quality does not improve. Therefore, if a certain threshold value is exceeded, the M value is determined to be small. .

  If it is determined that the M value is to be decreased, it is set to be smaller than the current M value (step S403). Here, since M = 1 to 4, in step S403, the M value is made 1 or 2 smaller than the current value. On the other hand, if it is determined in step S402 that the M value is not decreased, the optimum M value is determined by determining the increase or decrease of the M value using the number of motion vectors generated for each search range input in step S401 ( Step S404).

  Subsequent to the processing in step S403 or step S404, the optimum M value determination unit 116 outputs the optimum M value determined in step S403 or S404 to the M value control unit 117 (step S405). By using the methods of steps S403 and S404, the accuracy of the obtained M value is improved, and as a result, improvement in encoding efficiency and improvement in image quality can be expected.

  Next, an operation of determining the increase / decrease in the M value from the occupation rate of the motion vector in the generated code amount per picture by the optimum M value determination unit 116 will be described with reference to FIG. In FIG. 5, when the occupation rate (ratio) of the generated code amount in the generated code amount per encoded picture is smaller than the predetermined threshold A, the motion vector in the generated code amount per picture Since the generated code amount is small, the M value can be increased.

  On the other hand, when the occupation rate is equal to or greater than the predetermined threshold A, the ratio of the generated code amount of the motion vector to the generated code amount per picture is large, and there is a possibility that the code amount allocated to other than the motion vector is insufficient. Therefore, the M value is reduced, the generated code amount of the motion vector is reduced, and the code amount is assigned to other than the motion vector.

  Thus, by considering the occupation rate (ratio) of the generated code amount of the motion vector in the generated code amount per picture, the M value is changed optimally in a timely manner even at a low bit rate. Improve efficiency and improve image quality.

  Conventionally, since several images are input, first, a motion vector is obtained, then an optimum M value is determined, and encoding is performed with the determined M value. growing. On the other hand, in the present invention, since the next M value is determined from the image information of the previous M value using the correlation of moving images, the number of times of obtaining the motion vector is always one, Since the amount can be reduced, it is considered to be effective for broadcasting equipment that requires real-time processing. In addition, since it is possible to obtain an optimum M value with high image quality even at a low bit rate, it is considered to be effective for low rate devices such as mobile devices and recording devices (long-time mode).

  The present invention also includes a program that causes a computer to realize the functions of the above-described embodiments. This program may be read from a recording medium and loaded into a computer, or may be transmitted via a communication network and loaded into a computer. In addition, the present invention can be widely applied to a moving picture encoding apparatus using an inter-picture prediction encoding scheme, and can also be applied to an inter-picture prediction encoding scheme between fields.

It is a block diagram of one embodiment of the present invention. 2 is a flowchart for explaining the operation of a motion vector information detection unit in FIG. 1. It is explanatory drawing of an example of how to obtain | require the search range and motion vector for each M value in the motion vector information detection part in FIG. 2 is a flowchart for explaining an operation of an optimum M value determination unit in FIG. 1. FIG. 10 is a relationship diagram between an occupation rate of a generated code amount of a motion vector and an M value. It is explanatory drawing of an example of the search range of the motion vector by a conventional apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image input part 102 Image memory part 103 Subtractor 104 DCT part 105 Quantization part 106 Variable length coding (VLC) part 108 Code amount control part 109 Inverse quantization part 110 Inverse DCT part 111 Adder 112 Decoded image memory part 113 Motion compensation prediction unit 114 Motion vector information detection unit 115 Generated code amount count unit 116 Optimum M value determination unit 117 M value control unit

Claims (2)

A coded signal obtained by combining an I picture, a P picture, and a B picture obtained by coding a moving picture signal by three kinds of coding methods of intra-picture coding, forward inter-picture predictive coding, and bi-directional inter-picture predictive coding. Or between images that output an encoded signal that is a combination of an I picture and a P picture obtained by encoding a moving image signal using the two encoding methods of the intra-picture encoding and the forward inter-picture predictive encoding. A video encoding apparatus using a predictive encoding method, wherein an M value indicating an interval between adjacent I and P pictures in the encoded signal or an interval between two adjacent P pictures is variable. A video encoding device that performs the encoding,
Motion-compensated prediction means for obtaining a motion vector by performing motion compensation based on the reference image data obtained by locally decoding the encoded signal and the moving image signal;
Generated code amount counting means for determining the ratio of the generated code amount of the motion vector determined by the motion compensation prediction means in the total generated code amount per encoded current picture;
The motion compensation prediction means receives the magnitude of the motion vector, the search range of the motion vector, and the predicted absolute value error as inputs, and when the currently encoded picture is the P picture, The search range and the motion existing in the search range for each of a plurality of different M values determined in advance on the basis of the magnitude of the motion vector of the M value and the search range used when the motion vector was obtained. Motion vector information detection means for obtaining the number of generated vectors and selecting an M value corresponding to a search range having the largest number of generated motion vectors as an optimum M value;
The ratio obtained by the generated code amount counting means is compared with a predetermined threshold value. When the ratio exceeds the threshold value, the M value selected by the motion vector information detecting means is further reduced. Determined as the final optimum M value for the next encoding, and when the ratio is equal to or less than the threshold value, the final optimum M value selected by the motion vector information detecting means is encoded next. A moving picture coding apparatus comprising: an optimum M value deciding unit that decides as an M value; and coding based on the final optimum M value by designating a picture to be coded next. When the currently encoded picture is the B picture, the motion vector information detecting means encodes the P picture immediately before the forward motion vector obtained when the B picture is encoded. It is determined whether or not there is a predetermined number or more within the search range defined by the optimum M value selected when performing the operation, and the forward motion vector obtained when encoding the B picture is the predetermined number When there are more than the above, the optimum M value selected when encoding the P picture immediately before is selected as the optimum M value as it is, and when there are not more than the predetermined number, the B picture is replaced with the optimum M value. 2. The moving picture encoding apparatus according to claim 1, further comprising means for selecting the calculated M value as an optimum M value for the next encoding.

Images