A method and apparatus for encoding a target image within a sequence of one or more digital motion video images using motion compensation. Displacement vectors are assigned to a plurality of blocks in the target image. First and second block groups are selected from the plurality of blocks, the first and second block groups having corresponding first and second displacement vectors assigned thereto. A first error measure associated with predicting the second block group using the second displacement vector is determined. A second error measure associated with predicting the second block group using said first displacement vector is determined. An error change is obtained from the first and second error measures. A merge-factor is obtained from the error change. The second block group is reassigned to the first displacement vector if the merge-factor is less than a merge-threshold. Alternatively, displacement vectors are assigned to a plurality of parent blocks in the target image. Each parent block is divided into a plurality of target blocks. A target block is selected, and a plurality of parent blocks are associated therewith. A plurality of error measures are calculated by successively applying the displacement vector assigned to each associated parent block to the selected target block. The displacement vector associated with the minimum error measure is assigned to the selected target block if the minimum error measure is below a threshold. The target image is encoded in accordance with the displacement vectors currently assigned thereto.

A detecting circuit detects a moving vector for each block. Corresponding to the moving vectors, a plurality of moving portions in one field are extracted. Thus, a moving vector of each moving portion is generated. In a field memory, a motion compensating process is performed corresponding to the moving vectors. A field difference between the input signal and a signal that has been movement-compensated is supplied to an Hadamard transform circuit. The Hadamard transform circuit divides the field difference signal into a plurality of frequency components. A non-linear circuit has noise components corresponding to individual frequency bands. An output signal of the non-linear circuit is supplied to an inverse-Hadamard transform circuit. The inverse-Hadamard transform circuit performs an inverse-Hadamard transform process. The resultant signal is subtracted from the input picture signal.

An image signal treatment apparatus prevents degradation of picture quality which can occur because of various interferences phenomena arising from causes such as a hanging dot and its motion. The inventive apparatus performs proper correlation by taking into account the motion components of a composite image signal during Y/C signal separation. The inventive apparatus separates the luminance and chrominance signals of the composite image signal according to the motion components with respect to the composite image signal during separation. The apparatus is provided with filters which output luminance and chrominance signals by filtering the composite image signal horizontally, vertically and temporally. Motion detection circuitry outputs detection coefficients by detecting vertical and horizontal motion of the composite image signal. Mixers are provided to mix the luminance and chrominance signals outputted from the filters according to the detection coefficients of the respective motion detection circuitry. Adders add respectively the luminance and chrominance signals from the mixers, and amplifiers amplify and output respectively the luminance and chrominance signals outputted from the adders.

A system for processing 3-D multiple-view images which can render an image of a 3-D object as seen from any arbitrary viewpoint based on the 3-D multiple-view images, provided with an image acquisition unit for acquiring 3-D multiple-view images of a 3-D object as seen from a plurality of viewpoints, an image analysis unit for analyzing 3-D multiple-view images comprised of a plurality of images and finding the layered feature trajectory information comprised of trajectories linking identical points of the 3-D object, an image interpolation unit for rendering by interpolation an image of the 3-D object as seen from a specified (by user) viewpoint based on the layered feature trajectory information, an image compression unit for approximating the shape and color and texture information for the individual trajectories of the layered feature trajectory information for compression coding, a memory unit for storing the coded layered feature trajectory information, and an image reconstruction unit for reconstructing the coded layered feature trajectory information read from the memory unit to the original layered feature trajectory information and applying it to the image interpolation unit.

A scene-change detector is disclosed which detects a scene-change of a moving image composed of a plurality of frames. In the apparatus, a moving image is encoded based on the correlation between frames or inside a frame. The energy of an estimation error between frames in the case of encoding is operated by image elements. The energy is totaled by frames, and when the totaled value is larger than a specified value, a signal expressing a scene-change is output.