磨粒轮廓特征变换研究

Study on Translation for the Profile Feature of Wear Particles

本文提出了一个磨粒轮廓特征变换的方法,该方法解决了磨粒图像的平移、旋转和大小不变性,同时,该变换所得的结果能较为全面地反映了磨粒的轮廓特征,这使磨拉识别具有较高的准确率成为可能.文中给出了一个利用此变换方法所得的轮廓特征作为神经网络输入的磨粒识别实例.

This paper describes a method on translation for the profile of wear particles, which solve sinvariabiliry on displacement, rotation and scale of wear particles image. The results given by this method can reflect accurately the profile features of wear particles, which is the base to raise the accuracy for the wear particles recognition. An example using the method to get the input of a neural network for wear particles recognition is given in this paper.