Abstract: In allusion to the rich color information of gastroscope images，make use of neighborhood color moment histogram and the primitive histogram after quantization and clustering in HSV color space，and then use PCA algorithm to reduce the dimensions to retrieve，at last use relevance feedback to make multi-retrieve.The comparison of the retrieval results show that the approach proposed in the paper is effective.

Key words: gastroscope images, neighborhood color moment histogram, PCA algorithm, relevance feedback

摘要： 针对彩色胃窥镜图像丰富的颜色信息，在HSV颜色空间量化聚类后，利用多邻域颜色矩直方图和原始的量化聚类直方图的方法，再运用PCA算法降维，进行检索，最后利用相关反馈技术进行多次检索。实验表明，利用该方法在各种性能评价指标下都具有较好的效果。

关键词: 胃窥镜图像, 邻域颜色矩直方图, PCA算法, 相关反馈