论文部分内容阅读
由于自然景物图象的灰度与边缘特征不稳定，基于灰度与边缘的匹配算法对于这一类图象难以生效。本文通过分析ＦｒａｃｔａｌＢｒｏｗｎｉａｎＭｏｔｉｏｎ（ＦＢＭ）的Ｗｅｉｅｒｓｔｒａｓｓ－Ｍａｎｄｅｌｂｒｏｔ随机分形逼近函数的频谱，给出了分形向量特征的定义和快速判别图象的ＦＢＭ区域的方法。在ＦＢＭ区域内采用该特征进行匹配能克服复杂自然景物图象中灰度与边缘特征的不稳定性。实验表明采用ＦＢＭ分形向量特征的匹配方法能够获得比较传统匹配方法，例如平均绝对差算法以及特征匹配法更高的匹配定位精度和匹配概率
Due to the instability of gray and edge features of natural scene images, gray-edge and edge-based matching algorithms are not effective for this kind of image. In this paper, by analyzing the spectrum of Weierstrass-Mandelbrot fractal approximation function of Fractal Brownian Motion (FBM), the definition of fractal vector feature and the method of fast discriminating the FBM region of the image are given. Using this feature to match in the FBM area can overcome the instability of gray and edge features in complex natural scene images. Experiments show that the matching method based on FBM fractal vector feature can obtain more traditional matching methods such as average absolute difference algorithm and feature matching method with higher matching accuracy and matching probability