摘要： 针对基于内容的视频序列检索匹配提出一种综合颜色灰度和纹理特征对分块图像进行描述的方法，根据颜色灰度特征引入灰阶顺序码的概念。结合Harr小波变换与Gabor变换提出一种新的纹理提取方案。针对广告视频序列检索匹配的应用提出逐级过滤检索匹配策略，明显减小了计算复杂度，提高了检索效率。

关键词: 图像检索, 广告视频序列匹配, 分块, Gabor变换

Abstract: For content-based video sequence retrieval and matching, this paper brings forward a description method to block-image combined by grey and texture feature, while proposes the conception of grey-gap-order to describe grey feature. It unites Harr wavelet and Gabor transform, and puts forward a new method to distill texture feature. In addition, it brings forward a steped-sift scheme for advertising-video retrieval and matching, which remarkably decreases complexity of computing and boosts the retrieval efficiency.

Key words: image retrieval, advertising video sequences matching, block, Gabor transform

中图分类号: 

• TP312