关键词: 信息处理技术, 特征融合, 范围相似分, 遗传算法

Abstract:

An image feature fusion method was proposed based on scope similarity scores minimization. Aiming at the uncertainty of image feature fusion, it converted the problem of fusing image features into that of minimizing similarity scores of restricted samples on line. From different feature space, the similarity scores between a query image and images in database were calculated, and then some of them were selected to construct a candidate image pool, the size of which was relative to the number of return images. For the images in candidate pool, AG algorithm was employed to select return images, the purpose of which is to minimize the total similarity scores of return images. By this way, the adaptive weights of different image features were gained, and similar images were returned. The experimental results show that the proposed method is competitive.

Key words: information processing technology, feature fusion, scope similarity score, AG algorithm