基于纹理分析的表面粗糙度等级识别

提出了一种利用图象纹理分析技术进行机械加工表面粗糙度检测的非接触检测方法，该方法首先根据统计方差对待测工件的表面粗糙度进行粗分类，然后，利用基于Ｇａｂｏｒ滤波器的纹理分类器，识别待测工件表面粗糙度等级。该新方法可简单、快速地实现表面粗糙度等级的自动识别，而且对图象旋转具有不变性，由于其纹理分类器的参数少，并且新方法成本低，参数标定方便，因而便于现场检测，如果与机床的控制系统相连，还可以实现加工的实

Surface Roughness Detection Based on Texture Analysis

With the growing emphasis of industrial automation in manufact uring, vision techniques play an important role in many applications. Since diff erent surfaces have different textures, the techniques of texture analysis can b e used for the recognition of surfaces. In this paper, a novel non-contacted ap proach to measure the roughness of machined surfaces based on texture analysis t echniques is presented. When using Gabor filters, It is more complex to classify multiple textural images than to distinguish the texture between two images. Ac cording to other related paper and our experiments, the surface of a measured sp ecimen can be classified coarsely according to its gray-level variance. Then, t he roughness of the surface can be detected using Gabor filters. We present the method of designing the filters and the experiments show better results as well. The approach can detect the surface roughness automatically and quickly. It is invariant to rotation, and has fewer classifiers. Furthermore the cost of the de vice for implementing the approach is low and the parameters can be set easily. If the system is connected with the control system of a machine, we can realize real-time close looped control of the machining procedure.