基于图像局部投影特征的目标匹配算法(英文)

Matching Algorithm Based on Features of Sectional Projection of Images

鉴于传统的灰度相关匹配和基于边缘特征的匹配方法抗噪能力和抗几何失真能力较差,将图像熵和投影特征的概念结合引入到图像匹配中,定义了图像局部投影熵,并由此提出一种新颖的、基于局部投影熵差的图像匹配方法。由于局部投影熵反映的是图像的局部统计特征,因此基于局部投影熵差的图像匹配方法具有较好的抗几何失真能力。图像熵结合了投影特征后,提高了抗噪能力和在强光照条件下的匹配能力。结合采用分块、序贯检测及分层搜索等技术,进一步减少了计算量,实验结果表明这是一种简单而行之有效的图像匹配方法。

Traditional image matching methods, which are the gray level relating matching algorithm and the edge-based features matching algorithm, are not specialized in anti-geometrical distortion and anti-noise capability. The information entropy that was combined with the projection features and an image local projection entropy was defined. A novel image matching method based on the partial projection entropy difference was proposed. This image matching method has very good anti-geometrical distortion properties. Thanks to the combination of the image entropy and image projection feature, the anti-noise capability was raised and good matching result can be obtained in high light conditions. In addition, some methods are adopted such as dividing image to some blocks, series testing and multi-resolution searching techniques and the calculating amount is deduced. Experimental results show that this algorithm can work effectively.