基于图像处理和K近邻算法的示温漆判读方法  

Interpretation Method of Thermal Paint Based on Image Processing and K-nearest Neighbor Algorithm

作　　者：胡明 葛俊锋[2] 薛秀生 张宇 高佳祺 张泽鹏 HU Ming;GE Jun-feng;XUE Xiu-sheng;ZHANG Yu;GAO Jia-qi;ZHANG Ze-peng(AECC Shenyang Engine Research Institute,Shenyang 110015,China;School of Artificial Intelligence and Automation,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]中国航发沈阳发动机研究所,沈阳110015 [2]华中科技大学人工智能与自动化学院,武汉430074

出　　处：《航空发动机》2021年第6期80-84,共5页Aeroengine

基　　金：航空动力基础研究项目资助。

摘　　要：为了提高示温漆温度判读的精度和效率,采用图像处理和K近邻算法进行示温漆自动判读。对样板图像颜色空间进行转换,提取颜色特征,建立判读模型;采用基于邻域颜色相似性的多尺度分割算法对试验件图像进行分割和颜色空间转换,提取颜色特征。通过采用基于K近邻算法和颜色分布特征构建的温度判读模型,在KN3A示温漆矩形样板上进行了温度判读,结果表明:当允许误差在±10℃以内时,像素点温度判读准确率达到96%,区域温度判读准确率达到100%;在KN8蝶形样板上进行了温度判读,结果表明:热电偶附近像素点的温度值与热电偶温度值相比较,其误差在±10℃以内。In order to improve the accuracy and efficiency of temperature interpretation of thermal paint,image processing and K-nearest neighbor algorithm were used for automatic interpretation of thermal paint. The color space of the sample image was transformed,the color features were extracted and the interpretation model was established. The multi-scale segmentation algorithm based on neighborhood color similarity was used to segment the test piece image and transform the color space to extract the color features. The temperature interpretation model was constructed based on K-nearest neighbor algorithm and color distribution feature. The temperature interpretation was carried out on the rectangular sample plate of KN3 A thermal paint. The results show that when the allowable error is within ±10 ℃,the accuracy of pixel temperature interpretation is 96%,and the accuracy of regional temperature interpretation is 100%. The temperature interpretation was carried out on the KN8 bow-tie sample plate. The results show that the error between the pixel temperature near the thermocouple and the thermocouple temperature is within ±10 ℃.

关 键 词：示温漆 图像处理 K近邻算法 颜色分布 温度判读 

分 类 号：V219[航空宇航科学与技术—航空宇航推进理论与工程]