Abstract: In a method for identification of arborescent structures detected in digital images by starting from model arborescent structures, a matching operation consists in comparing an object tree with a model tree. The method involves computation of the distance between the object tree and the model tree in accordance with a so-called Selkow iterative method which utilizes predefined cost values for elementary operations of label-changing, insertion or destruction of sub-trees in the descriptive lists of the trees to be identified. Starting from all the steps of computation of the distance measurement, the elementary operations which have served to arrive at the final distance are recorded and identification is obtained by label changes of the sub-trees reduced to branches.

Abstract: A differential coding-decoding system, in which the coder comprises a forecaster delivering, on the basis of a reconstructed value corresponding to a coded quantity, a forecaster P for the following quantity X to be coded, a forecast error quantizer delivering the difference d=X-P, and a forecast quantizer, the outputs of which are connected to a code allocator, the allocated code word transmitted for a quantity X to be coded being related, first to the forecast error quantization level E.sub.i formulated by the forecast error quantizer, and, second, to the quantized forecast P.sub.j formulated by the forecast quantizer; the codes allocated to successive forecast error zones corresponding to forecasts P.sub.j are always assigned in the same order, when j varies from minimum to maximum forecast error zones.

Abstract: In the method, an image is analyzed by a television camera, is then submitted to differential coding with a "temporal inter-fields prediction", between fields of the same image, with the exception of one initialization line which is coded with a "uni-dimensional space prediction". The transmitting device and the receiving device of the transmission system each comprise an image memory which stores the two fields. They also comprise means for calculating the prediction, shown in the figure, and operating according to two types of the method. A first device calculates a uni-dimensional space prediction, and a second device calculates a temporal inter-fields prediction. A pointer selects the space prediction during initialization and the temporal prediction during the remainder of the transmission.