利用改进遗传算法的可靠性匹配方法

A Reliability Matching Method Based on an Improved Genetic Algorithm

提出了一种基于改进遗传算法的可靠性匹配方法。首先采用四阶中心B样条小波对实时图和基准图进行小波分解和边缘增强。然后,在粗尺度上利用改进的遗传算法对小波边缘增强图进行匹配,快速搜索并保留局部极值点作为候选匹配点,保证正确匹配点可以包含在候选匹配点中。对实时图进行旋转校正,利用候选匹配点处的局部灰度特征提取二值边缘图,再进行相关匹配,筛选出正确匹配点。最后在细尺度上获得精确的定位点。实验结果表明,其匹配概率比基于小波二值边缘提取的相关匹配算法提高了9.5%。

A reliability matching method based on an improved genetic algorithm was presented. The real-time image and the referenced image were decomposed to wavelets pyramid and edge-enhanced with the fourth order center B spline wavelets. The matching was done based on the improved genetic algorithm in coarse scale. The improved genetic algorithm can find the local extreme as the candidate matching points to ensure the true matching point was included. The real-time image was corrected with rotation. The binary edge images were obtained based on the local gray feature around the candidate matching points. The correlation matching was implemented again and the maximal peak value was chosen as the true matching point. Then the precise matching point was obtained in fine scale. Experimental results show that the matching probability of the method is increased 9.5% than that of the method based on the wavelets binary edge extraction.