AbstractThis paper describes a method in which three kinds of stereo matching methods based on different image features are integrated by a cooperative processing to realize a matching search system applicable to a wide range of objects. In other words, the point‐based matching using correlation, the interval‐based matching using edges and the segment‐based matching using connected edges are combined. The module for dedicated execution of each algorithm is provided, and the modules execute the matching search in parallel. Whenever necessary, the partial result of the processing is exchanged to realize a complementary cooperation.Each module continues the processing for the part which can be processed without being assisted by other modules. Then, considering each local area of the image, the best‐suited algorithm is selected and applied automatically according to the situation. Each module evaluates the extent of confidence of its own processing, and the reliability score is attached when the result of processing is exchanged. The module receiving the data appropriately considers the received data into its own processing, according to the confidence. The disparity image generation module integrates the results of three matchings according to the extent of confidence. The method is applied to several stereo images, and the usefulness of the proposed method is demon