采用图像增强的图像特征点匹配算法

首先,利用直方图均衡化、基于照明-反射模型的同态滤波和基于Retinex理论的3种图像增强方法,对收集到的图像进行预处理;然后,对预处理后的图像应用SIFT,SURF和ASIFT算法进行图像特征提取和匹配.研究结果表明:经直方图均衡化处理的图像,用SURF提取的图像特征点匹配率较高;同态滤波预处理图像的方法,经SIFT算法提取图像特征点的匹配率较高.

Image Feature Point Matching Algorithm Research Using Image Enhancement

Firstly, use the histogram equalization, the homomorphic filter based on light and reflection model, and the image enhancement method based on the Retinex theory are applied respectively to preprocess the collected images; and then three algorithms of SIFT, SURF and ASIFT are used for the preprocessing images to extract and match the image feature. Research results show that: by the image histogram equalization processing, the feature points matching rate of SURF was used to extract images is higher; Homomorphic filtering image preprocessing method, by the SIFT algorithm to extract image feature points matching rate is also higher.