CAJ | 학술논문

采用机器视觉技术，针对新疆无核白葡萄干的颜色特征进行分级研究，提取 HIS 颜色图像，采用中值滤波法滤波，采用最大类间方差法分割图像，采用形态学开运算去除二值图像中的伪目标区域，获得最佳二值图像。同时，分析色调灰度直方图和颜色矩均值直方图，确定采用 H、S、I 分量颜色的一阶矩、二阶矩、三阶矩作为特征值建立BP 神经网络的色泽分级模型，分级准确率最高为96．42％。
채용궤기시각기술，침대신강무핵백포도간적안색특정진행분급연구，제취 HIS 안색도상，채용중치려파법려파，채용최대류간방차법분할도상，채용형태학개운산거제이치도상중적위목표구역，획득최가이치도상。동시，분석색조회도직방도화안색구균치직방도，학정채용 H、S、I 분량안색적일계구、이계구、삼계구작위특정치건립BP 신경망락적색택분급모형，분급준학솔최고위96．42％。
The object of this study is to grade color features of xinjiang thompson seedless raisins ,extracting the HSI im-age,median filtering method is used to filter the noise ,OTSU method is used for image segmentation ,morphological open operation is used to remove the false target area in the binary image ,so that we can get the best binary image .Analysising the tonal cumulative gray histogram and color moments histogram is used to determine the first moment ,second moment and the third moment of H,S,I color components,color moments are as characteristic values to establish the BP neural network classification model ,the highest classification accuracy is 96 .42%.