A circuit for implementing a full search block matching algorithm for coding video signals sequentially receives pixel values from a block of pixels in a current video frame and sequentially receives pixel values from a search area of a previous video frame. The circuit includes a plurality of processors for calculating in parallel a group of error functions corresponding to a group of positions of the block in the search area. The error functions are compared for determining the block position having the minimum error function.

A motion vector detecting apparatus for detecting motion vectors based on a current frame data and the preceding frame data. In the motion vector detecting apparatus, first and second memories are provided in order to store the current frame data and the preceding frame data. Additionally, a number of first cache memories, a number of second cache memories, a control circuit, and a motion vector calculation circuit are provided. The control circuit selects the input cache memories into which pixel data are to be written, and the output cache memories, from which pixel data are to be read out, from the first and second cache memories, so that the input cache memories and the output cache memories have no redundant cache memory. Pixel data corresponding to the detection image blocks stored in the first memory and pixel data corresponding to the search image blocks stored in the second memory are sequentially written in the selected input cache memories of the first and second cache memories. Simultaneously, the pixel data corresponding to the detection image blocks and the search image blocks are sequentially read out from the selected output cache memories, and are sequentially supplied to the motion vector calculation circuit.

In an image compression system using a typical image compression scheme, a pointer array is provided to point to each of the many MCUs in a compressed image file. From all the blocks of an image, a subset of the blocks is selected as a virtual image. The virtual image is edited, and each edited block is compressed into an edited block. The edited block is compressed into an edited MCU and placed in an edited block region, and the pointer to the original MCU is changed to point to the new MCU. In this way, the pointer array can be modified to perform an Undo operation. An edge table is provided to hold values where each value, when combined with the differential value for a block on the edge of the virtual image, provides an absolute value for the block without reference to blocks beyond the edge of the virtual image. The entries in the edge table are determined from the compressed MCUs without the blocks being fully decompressed. More than one edge table can be provided. In an image editor, a virtual image is decompressed from a compressed image, the virtual image is processed, and recompressed. The recompressed, edited blocks are then placed in an edited block memory. In an alternate embodiment, values to be combined with a differential value are held in an offset table for all the selectable blocks.

In a motion estimator which adds small increments (P-2, P-4) to a prediction vector (P) to obtain a number of candidate vectors (2,4) from which an output vector is selected, a first increment (P-2) in a first direction has a different magnitude from a second increment (P-4) in a second direction.

Method of coding images converted into digital video signals in accordance with a format having three-dimensional blocks of the signals, wherein the method includes the following preliminary steps: (a) estimation of a principal movement from image to image for associating a displacement vector with each image, with respect to the preceding image, said principal vector being the vector for which the image-to-image difference is minimum, and; (b) scan conversion with a view to defining the format of the three-dimensional blocks by dividing the sequence of video signals corresponding to images into groups each corresponding to N consecutive images and by defining within each of these groups the three-dimensional blocks based on M lines, and P pixels per line on the one hand, in the image plane and, on the other hand, N successive planes corresponding to n images of the group, the N two-dimensional blocks of M lines by P pixels which compose each three-dimensional block of one and the same group being spatially shifted from one image to the next by the said displacement vector estimated for each image.