基于视觉的坐便器清洁度检测方法应用研究  

Application research of visual-based method for detecting toilet cleanliness

作　　者：彭勇 张华 高延峰 PENG Yong;ZHANG Hua;GAO Yanfeng(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学机械与汽车工程学院,上海201620

出　　处：《上海工程技术大学学报》2024年第3期334-340,348,共8页Journal of Shanghai University of Engineering Science

基　　金：国家自然科学基金资助(61763030);国家重点研发计划“智能机器人”重点专项资助(2018YFB1305304)。

摘　　要：目前二便智能护理机器人只在排便结束后对坐便器斗进行一次恒定大水量冲洗,没有相应清洁度检测环节,导致残存的粪便容易造成空气异味和细菌滋生,甚至疾病传染。针对于此,提出基于视觉的两种清洁度检测方法:一种是通过计算污点像素占比来进行坐便器清洁度的评估,另一种是采用图像模板匹配的方法来检测坐便器的清洁度。试验结果表明:采取自适应阈值分割来计算污点像素占比时能够很好地克服背景阴影的影响,且能够准确地区分出不同污染程度的照片组,达到便斗清洁度检测的目的。At present,intelligent nursing robots for secondary defecation only rinse the bucket with a constant amount of water after defecation,without corresponding cleanliness testing.The remaining feces can easily cause air odor and bacterial growth,and even spread diseases.In response to this,two visual based cleanliness detection methods were proposed:one is to evaluate the cleanliness of the toilet by calculating the proportion of dirty pixels,and the other is to use image template matching method to detect the cleanliness of the toilet.Experimental results show that by using adaptive threshold segmentation to calculate the proportion of dirty pixels,it can effectively overcome the influence of background shadows and accurately distinguish photo groups with different levels of pollution,thus achieving the goal of detecting the cleanliness of the toilet.

关 键 词：清洁度 图像处理 图像相似度 自适应阈值分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]