基于Relief算法的故障图像识别与匹配方法

A Method of Malfunction Images Recognition and Classification Based on Relief

针对图像识别与匹配问题,提出一种基于Relief算法的新方案。利用k最近邻的Relief算法选择表达图像的最优特征子集,再利用基于最小距离分类器的模板匹配技术,实现故障图像的识别。在特征提取时,改进了对传统的灰度共生矩阵的选取。实验证明,该算法正确识别图像故障与否的概率能到达90%左右,提高了特征选择的有效性,完全满足实际应用中的需要。

A new method based on Relief algorithm was proposed to solve image identification problem.K neighbor Relief algorithm was used to choose the optimal feature subset of images.Minimum distance of template matching was used in image recognition.This method improved the selection of gray co-occurrence matrix.Experiments indicated that the current recognition probability can reach 90%,this algorithm not only resulted a substantial improvement in the feature selection,but also met the needs of practical engineering.