融合色彩增强与投影展平的蚕茧疵点面积计算方法  

Calculation Method of Cocoon Defect Area by Fusion of Color Enhancement and Projection Flattening

作　　者：虞旭锋 孙卫红 邵铁锋 梁曼[1,2] 高明辉 Yu Xufeng;Sun Weihong;Shao Tiefeng;Liang Man;Gao Minghui(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China;Cocoon and Silk Quality Inspection Technology Institute,China Jiliang University,Hangzhou 310018,China;Tai’an Textile Fiber Inspection Institute of Shandong Province,Tai’an Shandong 271000,China)

机构地区：[1]中国计量大学机电工程学院,杭州310018 [2]中国计量大学茧丝绸质量检测技术研究所,杭州310018 [3]山东省泰安市纺织纤维检验所,山东泰安271000

出　　处：《蚕业科学》2024年第5期417-425,共9页ACTA SERICOLOGICA SINICA

基　　金：浙江省公益技术应用研究项目(LGG20E050014);国家市场监管总局科技计划项目(S2021MK0217)。

摘　　要：在蚕茧分级过程中,常以疵点面积作为评价指标。为计算存在颜色缺陷的蚕茧的表面疵点面积,文中提出了一种融合色彩增强与投影展平的蚕茧疵点面积计算方法。首先,对亮度增强后的原始图像V通道进行阈值分割及矩形拟合,获取图像ROI(region of interest)区域;然后,在带颜色恢复的多尺度Retinex(multi-scale Retinex with chromaticity preservation,MSRCP)算法中添加完美反射白平衡算法,配合多尺度快速引导滤波器,提高疵点区域的对比度;最后,通过投影展平将投影面积转化为实际曲面面积。实验结果表明,文中算法计算所得面积与人工测量面积的误差范围在3.8%~8.7%,最大误差面积为0.07 cm^(2),满足次下茧分级的精度要求。In the process of grading silkworm cocoons,the defect area is often used as an evaluation metric.To calculate the defect area of cocoons with color defects,a method combining color enhancement and projection flattening for defect area calculation was proposed in this study.Firstly,threshold segmentation and rectangle fitting were applied to the V channel of the brightness-enhanced source image to obtain the ROI(region of interest)region of the image.Secondly,the perfect reflection white balance algorithm was added to the MSRCP(multi-scale Retinex with chromaticity preservation)algorithm to match the multi-scale fast guide filter,thereby improving the contrast of defect areas.Lastly,the projected area was converted into the actual surface area through projection flattening.Experimental results showed that the error range between the area calculated by this method and manually measured area was between 3.8%and 8.7%,with a maximum error area of 0.07 cm^(2),meeting the accuracy requirements for grading inferior cocoons.

关 键 词：疵点面积 图像增强 引导滤波 MSRCP算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]