论文部分内容阅读
采用计算机视觉采集珍珠图像,通过一系列的预处理后,转化到极坐标系下,利用计算前8组F(h)(傅里叶系数)数值作为每种典型形状特征面的特征参数,然后运用模糊模式识别的方法对每一幅图像的珍珠形状进行有效判别。通过对多视角得到的图像特征面的寻找与比较判别,实现珍珠形状的判别分类。实验结果表明,分选最大误判率为33.3%。
The pearl images were collected by computer vision and transformed into the polar coordinate system through a series of preprocessing. By using the first eight groups of F (h) (Fourier coefficients) as the characteristic parameters of each typical shape feature surface, The method of fuzzy pattern recognition is used to effectively distinguish the pearl shape of each image. Through the multi-angle image feature plane to find and compare discriminant, to achieve the classification of pearl shape discrimination. Experimental results show that the maximum false positive rate of 33.3%.