摘要
提出了一种基于计算机图像处理技术的畸形秀珍菇识别方法。研究根据正常和畸形秀珍菇的形状特征,通过统计性分析,提取了分形维数、相对位移、菌盖圆形度、菌盖形状因子、菌盖凸性率、菌盖偏心率、菌柄弯曲度等7个特征参数。通过逐步回归筛选出分形维数、相对位移、菌盖偏心率、菌柄弯曲度等4个特征变量,并将这4个特征变量作为输入向量,采用支持向量机模式识别方法建立畸形秀珍菇判别模型,模型的独立样本预测集实测值识别率达96.67%。研究表明,利用机器视觉技术能很好地识别畸形秀珍菇,研究方法和结果为实现秀珍菇的在线分选提供技术支持。
An identification method was developed to automatically recognize defect Pleurotus Geesteranus based on computer image processing technology.Seven feature parameters were extracted acording to investigation of shapes of Pleurotus Geesteranus,which were fractal dimension,relative length,roundness,shape factor,convexity of the pileus,aspect ratio,and crooked degree of the stipe.Subsequently,four feature parameters were further extracted by stepwise linear regression,which were fractal dimension,relative length,aspect ratio,and crooked degree of the stipe.Finally,support vector machine classifier was employed to build discrimination model,where four feature parameters selected were used as inputs vector.Recognition rate of model was 96.67%,when discrimination model was tested by some independent samples in the prediction set.This study demonstrates that it is feasible to identify defect Pleurotus Geesteranus using machine vision technique and provide technical support for on-line grading of Pleurotus Geesteranus.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2010年第10期350-354,共5页
Transactions of the Chinese Society of Agricultural Engineering
基金
江苏省青蓝工程资助项目
江苏省自然基金(BK2007087)
关键词
识别
图像处理
支持向量机(SVM)
秀珍菇
identification
image processing
support vector machines（SVM）
Pleurotus Geesteranus
