珍珠形状的计算机视觉识别  

Pearl Shape Recognition Based on Computer Vision

作　　者：李革[1] 李斌[1] 王莹[1] 余智[2] 李兢[3] 赵华勇[1] 

机构地区：[1]浙江理工大学机械与自动控制学院 [2]浙江工商大学现代教育技术中心 [3]浙江理工大学科技处

出　　处：《农业机械学报》2008年第7期129-132,共4页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金资助项目(项目编号:50575211);浙江省科技厅新苗人才计划项目(项目编号:2007R40G2060035)

摘　　要：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%。Shape classification is an important step to pearl classification. After a series of preprocessing, the pearl image obtained by computer vision was transformed to the polar coordinates, and then eight F（h） （Fourier coefficient） values was computed as each kind of typical shape characteristic parameters. Subsequently the utilization of fuzzy pattern recognition method realized the shape effective distinction of each pearl image. Finally, the pearl shape distinction classification was realized by seeking and comparison of characteristics image through multi-angles of view. The experimental results indicate that the largest error judgments rate of the proposed distinction method is 33.3 %.

关 键 词：珍珠 计算机视觉 傅里叶变换 模糊模式识别 

分 类 号：S126[农业科学—农业基础科学] TP391.41[自动化与计算机技术—计算机应用技术]