支持向量机在注塑产品图像预处理中的应用

分析了现有中值滤波器算法在注塑产品图像预处理过程中应用时存在的问题,结合注塑产品图像的特点,提出了一种新的多级中值滤波器算法;通过训练基于支持向量机的像素点分类器,实现了对图像中像素点类型的判别,进而根据其自身特性的不同,有的放矢地选择相应的滤波处理方法.该算法既有效地抑制了注塑产品图像中存在的噪声污染,又提高了滤波算法保护图像中边缘细节信息的能力,更好地满足了注塑产品图像预处理的要求.仿真研究证明了该算法的有效性.

Application of SVM to Image Pre-processing of Injection-Molding for Plastic Products

Analyzing the problems found in the conventional median filters applied to the image pre-processing of injection-molding for plastic products,a novel multistage median filter is proposed to suit well the images plastic products.By the training of a SVM-based sorter,the type of pixels in image can be distinguished from each other so as to enable the new filter to choose properly a filtering algorithm in accordance to the type of pixels.In this way the SVM-based multi-stage median filter offers an excellent capability for edge detail protection with noise suppressed efficiently,thus satisfying the requirements of the image pre-processing of plastic products.Simulation tests have verified the effectiveness of the new filter.