Abstract: Integrating multiple features content-based image retrieval can overcome the problems of single feature，but how to organize these features and feature representation methods are difficult in image retrieval.An automatic feature weight assignment approach based on genetic algorithm is proposed.First the problem of weight assignment is changed into optimization problem，then genetic algorithm is used for finding the optimization weight in order to get the best retrieval results.The experimental results show that the recall and precision of this proposed approach is better.This approach is robust to various kinds of features and feature representation methods，and it can get the best feature combination for image retrieval.

Key words: content-based image retrieval, genetic algorithm, feature weigh

摘要： 综合多种特征的图像检索可以克服使用单一特征带来的问题，如何组织这些特征使其发挥最佳作用是基于内容图像检索中的一个难点。提出了一种基于遗传算法的多特征权重自动赋值方法，首先将图像检索中特征权重调整问题转换为优化问题，进一步利用遗传算法求取权重的最优解，使检索效果达到最好。实验表明，该方法查全率和查准率较高，并且具有较强的鲁棒性和推广能力，对于各种特征和特征表示方法，都能够提供最佳的组合。

关键词: 基于内容的图像检索, 遗传算法, 特征权重