基于分形维数的柑橘形状与光滑度的机器视觉分级  

Citrus Fruits Grading by Shape and Smoothness Based on Machine Vision and Fractal Dimension

作　　者：曹乐平[1] 温芝元[2] 陈理渊[3] 

机构地区：[1]湖南生物机电职业技术学院教务处,湖南长沙410127 [2]湖南农业大学理学院,湖南长沙410128 [3]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《测试技术学报》2009年第5期407-411,共5页Journal of Test and Measurement Technology

基　　金：湖南省自然科学基金资助项目(2007JJ6129);湖南省农业厅重点课题资助项目(200704A);湖南省教育厅科学研究项目(06D059)

摘　　要：为了实现柑橘形状与果皮光滑度的自动分级,研究了柑橘分形维数与柑橘形状和果皮光滑度间的对应关系.对机器视觉系统采集的柑橘花萼面和侧面的图像设置蓝色分量阈值去除背景后,经图像二值化、四连通化边界跟踪和边界细化操作,提取柑橘边界周长和柑橘区域面积,利用周长-面积方法计算柑橘分形维数,以此度量柑橘形状和果皮光滑度.3组仿形样本的分析表明,分形维数随形状和表面光滑度的变化而变化,说明分形维数既概括了柑橘外部形状特征,同时又包含了果皮光滑度信息.用10个柑橘检验样本的分形维数评定柑橘等级,其结果与人工感官判断结果完全吻合,这表明:分形维数能对柑橘形状和果皮光滑度进行分级.The relation of fractal dimension to fruit shape and pericarp smoothness are investigated to achieve auto grading of citrus fruits. After fruit calyx and profile image backgrounds are removed with threshold values, these images are converted to binary images. With boundary tracing and thinning, the perimeter and area of fruit region in the image are obtained to calculate the fractal dimension of the citrus fruit, and to determine fruit shape and smoothness of pericarp. Test results of 3 profile modeling groups show that the fractal dimension varies with fruit shape and smoothness, which implies the fractal dimension contains the information of fruit shape and pericarp smoothness. Ten samples were graded with fractal dimension, and the grading results were consistent with that of manual work.

关 键 词：分形维数 机器视觉 柑橘 形状与光滑度 分级 

分 类 号：TN911.73[电子电信—通信与信息系统]