No Title




6 and the range of movement correction for a block has a rate of ±1 picture element per frame in the horizontal direction and ±1 line per body within the vertical path as within the case of the motion compensation range described in conjunction with FIGS. Then, so as to appropriate the movement inside this range for this block, sixteen predictions from x1 to x16 are essential for 9 prediction vectors from which is mainly used for a stationary picture in the earlier frame, , , . 3B. For example, when the vector is used for prediction, a block to be predicted as x3, x4, x14 and x5 which is obtained by shifting zero image element, i.e. no movement, in the horizontal path learn free here and one line downward, e.g. a block including a11, a12, a21 and a22 as shown in FIG. When vector (1, -1) is used for prediction, computation is carried out for x4, x12, x15 and x16. Thus 16 prediction signals are necessary for the above-mentioned motion compensation vary. When it's the minimum, the prediction operate comparator 8--8 selects and holds the prediction function provided through line 108-5b right now as a brief optimum prediction perform, but when the outcomes of judgment is "no" the comparator 8--8 holds the previously determined temporary optimum prediction operate as it is. The momentary optimum operate is at all times equipped to a prediction choice circuit 8-9 which delivers the detected optimum prediction operate, that is, a motion vector onto the line 108 in accordance with a timing sign similar to the divisions of the blocks and equipped from the sequence 8-5 over the line 108-5b.

Where the transmission path has a high transmission speed so that it is attainable to assign several bits to 1 picture element on a median, the movement vector info occupies solely a portion of everything of information transmitted. However, the place the transmission pace is low, the motion vector info occupies a considerable portion of the entirety of knowledge transmitted, for instance, more than 50%. An instance of the movement vector info measured as regards to an actual image image, is talked about in a Hirano et al paper entitled "Investigation relating to the Block Size in the movement compensated inter-frame Coding", lecture No. 697 introduced before the general assembly of the Institute of Electronics and Communication Engineers of Japan, communication part, 1980. According to this paper, the entropy of movement vector utilized as a measure of the movement vector info is about zero.1 bit/picture element the place a block constituted by a plurality of image components has a measurement of 4 lines×8 picture components.

One of four parallel image factor outputs is equipped by way of line 108-6a to a subtractor 8-12a, proven in FIG. The position of an image element processed by the arithmetic unit 8-1 might be referred to as hereinafter picture element level D1 as shown in FIG. 3A. Similarly, other arithmetic models 8-2 via 8-4 are used for computing image component factors D2 through D4, and 4 arithmetic units in whole are used in parallel for concurrent computations for one block. Line reminiscence pair 1-1 and 1-2 and one other line memory pair 1-3 more info here.. and 1-4 are saved and browse out alternately, so that the video signal which is rendered scan conversion i.e. converted into two-dimensional blocks, is output in a string of picture parts. One of two outputs from a line reminiscence pair in reading operation is selectively output by way of the multiplexor 1-5. The block-formation video signal on a line one hundred and one from the multiplexor 1-5 branches to the traces 101a and 101b, then is equipped to the delay circuit 2a and optimum prediction detector 8, respectively. The above-mentioned scan conversion is advantageous since the predictive coding circuit utilizing the two-dimensional blocks is simple.

Therefore, the reproduced image sign is transformed by the scan inverter circuit 14 again to the video signal in unique time series, then the decoded video sign is sent out on line 214. 10 is a block diagram showing an embodiment of the predictive decoding system in accordance with the present invention. In the determine, the video signal which has been coded by prediction by means of the predictive coding system of FIG.

This system is the simplest one for assuring a excessive coding efficiency even for a transferring picture. With this movement compensation inter-frame prediction system, for the aim of sustaining a high effectivity for various movements, many prediction functions are used. More notably, a plurality of prediction features similar to movements at various speeds and in numerous instructions are prepared beforehand, homepage and a prediction perform most adaptable for an truly occurring movement is used as an optimum function.

The sum of those two code lengths is taken into account to be an quantity of data generated when the actual prediction function is used. With regard to the opposite prediction functions, the sum of two code lengths is equally determined. A prediction function which is obtained when the results of comparison of the sum of the 2 code lengths shows a minimal is set as an optimum prediction operate, that is, a motion vector. It is feasible to make smaller the amount of the transmission data when the motion vector is set on this method than when the movement vector is decided through the use of only the prediction error information as within the prior artwork. The percentage of the codes necessary to represent this motion vector with respect to all of the transmission information varies significantly depending upon the transmission speed on the transmission path used.

For the predictive coding system utilizing motion compensation, the range of correction, i.e., the variety of vectors N, ought to preferably be as giant as attainable and the variety of image elements in a block be as small as potential. From this viewpoint, the tactic of in which necessary hardware is determined by the worth of N and a wide variety of motion compensation disadvantageously wants an increased hardware. In distinction, the tactic of mainly makes use of M arithmetic models repeatedly for N times, and the hardware requirement is virtually determined by the worth of M. Thus, it could be seen that the tactic of is more advantageous. The tackle of the prediction signal generator memory 8-11a navigate here for writing the above-mentioned supplementary picture components is ideally next to the addresses for storing picture elements that are essential for determining the optimum prediction function for the present block, for facilitating read/write control. For this function, A1 via A8 are written following the addresses for storing x1 by way of x16 as shown in FIG. The necessary storage capability of the prediction sign generator reminiscence 8-11a is expressed, utilizing symbols k, L, m and n, as (2m+k)×(2n+L) for determining the optimum prediction function of the present block, k×(2n+L) for supplementary picture parts for figuring out the optimum prediction perform of the following block, amounting to 2(m+k)×(2n+L) in complete.