Adaptive vision system using dual thresholding

(57) Summary book invention relates to the picture analysis method and system for identifying an object in a background by defining data space like a histogram, a 共発生 matrix, or a color space. Data space possesses two or more subspace, and subspace is chosen based on the way a histogram, a gray level value, or a color parameter crowds in each space, for example. At least one subspace is chosen. To at least one expression of candidate objects, a picture uses selection secondary space, multiplex search is carried out, and candidate objects have at least one predetermined attribute value in this case. An appropriate object is identified by comparing the regulation set of an appropriate object attribute value and candidate-objects attribute value which were included in the driver. The present invention includes recursive, repetition, and the parallel processing method. A method -- a colony -- it is used in very variegated industrial inspection technology including discernment of the paints element embedded at the discrete feature and polymer in calculation and a carpet.