基于最大互信息方法的机械零件图像识别

提出了应用最大互信息方法进行零件图像识别的方法,它利用图像的信息熵描述图像的特征,结合图像的颜色信息及局部形状信息,以互信息作为衡量两幅图像相似性的测度函数进行图像识别,弥补了直方图表达空间信息的不足。该方法既满足位置不变性,又能避免进行图像分割,从而避免了因图像分割引起的复杂计算,使算法容易实现。实验结果表明,该方法提高了零件图像识别的精度、稳定性和可靠性。

Machine Part Image Recognition by Using Maximization of Mutual Information

A new approach to the problem of machine part image recognition is proposed by using maximization of mutual information. The method applies entropy to measure image feature, combined with color information and local shape information, and uses mutual information as a new matching criterion between the images for image recognition. This method solves the problem that histogram algorithm can not represent the spatial information. This method not only has the feature of translation invariant, but also avoids image segmentation which may lead to a complex calculation, so it can be realized easily. The result shows that proposed approach is accuracy, stability, and reliability in the processing of machine part image recognition.