When the optimum prediction perform more info here.. is determined in actual time processing for a block including 32 image parts, for instance, it have to be carried out within 32 image component intervals. 7 and eight are block diagrams displaying intimately the optimum prediction detector 8 of FIG. 7, the output signal from the delay circuit 6 equipped through the line 106b to the arithmetic unit 8-1 is quickly saved in a prediction sign generator memory 8-11a inside the arithmetic unit 8-1. The block-formation input sign is provided via line 101b to a serial to parallel converter 8-6, and here 4 image components in a block, e.g.
Means for compressive coding of information, stated data containing a minimal of information representing stated optimum prediction function and data representing a prediction error delivered out of stated predictive coding means when said optimum prediction perform is used. The code compressor 7 has a data price matching perform which permits compressed info to be output at a continuing data fee, and it features a prediction error signal encoder 7-1 related between line 103a and a multiplexor 7-3, and a motion vector encoder 7-2 related between line 108b and multiplexor 7-3 as shown in FIG. An adaptive coding system based on claim four whereby stated code length of said variable word length code is minimal for said inter-frame prediction and increases because the norm of mentioned prediction perform increases. An adaptive coding system based on declare 2 whereby stated first analysis means evaluates mentioned prediction perform by encoding it into a variable word size code which has the minimal code size for the inter-frame prediction.
One of 4 parallel picture factor outputs is supplied via line 108-6a to a subtractor 8-12a, proven in FIG. The position of a picture component processed by the arithmetic unit 8-1 shall be called hereinafter picture component point D1 as shown in FIG. 3A. Similarly, other arithmetic models 8-2 via 8-4 are used for computing picture factor factors D2 via D4, and four arithmetic models in complete are used in parallel for concurrent computations for one block. Line memory pair 1-1 and 1-2 and another line reminiscence pair 1-3 and 1-4 are stored and skim out alternately, so that look at this site the video sign which is rendered scan conversion i.e. converted into two-dimensional blocks, is output in a string of picture components. One of two outputs from a line memory pair in reading operation is selectively output through the multiplexor 1-5. The block-formation video sign on a line one hundred and one from the multiplexor 1-5 branches to the traces 101a and learn more 101b, then is provided to the delay circuit 2a and optimum prediction detector eight, respectively. The above-mentioned scan conversion is advantageous for the rationale that predictive coding circuit utilizing the two-dimensional blocks is straightforward.
An instance of the tactic of detecting the amount of displacement, that is, a technique of figuring out an optimum prediction perform is disclosed in U.S. No. four,307,420 entitled "Motion-compensated interframe coding system", issued Dec. 22, 1981. According to this patent, prediction errors concerning varied prediction capabilities similar to vectors indicative of assorted actions are compared with one another in block items and a vector similar to a prediction function that gives the minimal prediction error is deemed as a motion vector consultant of an optimum prediction operate. With regard to the impact of movement compensation, it is stated that by applying the movement correction for a shifting image, the quantity of information can be reduced to about one-half of the inter-frame coding system, thus attaining considerable compression of data. An assumption is made here that the picture parts included in a block are disposed in a two-dimensional 2-by-2 array as illustrated in FIG.
5 is a block diagram showing intimately the scan converter 1 of FIG. 4A. In this embodiment, 2×2 two-dimensional blocks every consisting of two picture elements in the horizontal path and two strains within the vertical course, as shown in FIG. One block contains 4 image parts, e.g. a11, a12, a21 and a22, enclosed by broken line. In a common tv system, the display screen is scanned from left to proper within the order of, for example, a11, a12 , a13 and a14, and parts a21, a22, a23 and a24 are included on the succeeding scanning line.
Preferably, the movement vector data is coded right click this link into a variable word size code which has the minimum code size for the inter-frame prediction. Specifically, the code length of the variable word size code is minimal for the inter-frame prediction and increases because the norm of the motion vector increases. In the previous, a relatively excessive transmission speed was assumed so that it has been considered that the share of the motion vector data is negligibly small. Consequently, it has been thought-about that an optimum prediction which minimizes the amount of everything of data transmitted can be realized by detecting a motion vector which minimizes the amount of data representing a prediction error.
For the predictive coding system using motion compensation, the range of correction, i.e., the variety of vectors N, should preferably be as giant as potential and the variety of picture elements in a block be as small as attainable. From this viewpoint, the strategy of by which essential hardware is decided by the value of N and a extensive range of movement compensation disadvantageously wants an increased hardware. In distinction, the tactic of primarily uses M arithmetic items repeatedly for N times, and the hardware requirement is virtually determined by the value of M. Thus, it may be seen that the strategy of is more advantageous. The address of the prediction sign generator memory 8-11a for writing the above-mentioned supplementary picture parts is ideally subsequent to the addresses for storing image components that are necessary for figuring out the optimum prediction perform for the current block, for facilitating read/write control. For this objective, A1 through A8 are written following the addresses for storing x1 via x16 as shown in FIG. The essential storage capacity of the prediction signal generator memory 8-11a is expressed, using symbols k, L, m and n, as (2m+k)×(2n+L) for figuring out the optimum prediction operate of the present block, k×(2n+L) for supplementary image elements for figuring out the optimum prediction perform of the following block, amounting to 2(m+k)×(2n+L) in total.