Application Research of Computer Artificial Intelligence Technology in Image Search and Recognition System

In order to improve the data transfer rate and identification search efficiency of image recognition, a software driver interface design architecture based on FACADE is established in this paper. The PCI device is driven by MDL memory and the vector identification software is designed. Then, a background occlusion method based on microscopic search is proposed to find an efficient decision subset to replace the traditional decision set. This allows efficient retrieval of high-precision image matting. The comparison with the classical matting method proves that the proposed method can effectively overcome the occlusion problem when the foreground and background are highly similar. At the same time, it can effectively reduce the dimension of high-score image matting problem, so as to improve the accuracy of foreground occlusion extraction.