一种基于彩色特征点的对象查询方法

图像检索系统中对象的查询，主要采用分割技术来实现。虽然一些研究人员提出过利用光学特征的方法，但是只应用于灰度图像，不能对大量的彩色图像进行有效的查询。为此本文提出了一种基于彩色特征点的对象查询新方法。首先，利用Harris彩色点提取嚣从彩色图像中提取出一组能表征图像特征的彩色点；其次，利用特征点的梯度、角度不变性构造检索策略来实现对象的查询。实验表明，该方法对于对象的光照变化和几何变化有较强的鲁棒性。

Object-Based Queries Using Color Character-Point

Object-based queries in image retrieval system usually imply segmentation. Some researchers have brought a new method forward that is interested in optics. However it is not optimal.since it is limited to gray value images. This method cannot solve color images. For this reason,this paper proposes a new solution based on color character-point. Firstly.it requires using the Harris color detector to extract some points from images. These points are good enough to represent feature of mages; Secondly .using grads and angles invariants of the characterization of points to implement queries. The result of the experiment shows that this solution is robust to object perturbations like viewpoint or illumination changes.