基于紫外成像技术的彩色光斑映射识别算法研究

Research on Color Spot Mapping Recognition Algorithm Based on Ultraviolet Imaging Technology

紫外成像检测是一种有效的非接触式放电检测方法,紫外放电图像中彩色光斑的准确分割直接关系到其后续诊断的准确性。因此,提出了一种基于灰度差异的紫外图像彩色光斑区域识别分割方法,旨在准确提取彩色的紫外光斑区域。首先,利用紫外成像仪采集了电气设备紫外放电视频,其中放电光斑区域颜色涵盖了目前紫外成像仪常用的色彩;其次,改进了紫外视频的处理方法,利用高斯函数对紫外图像进行预处理,将光斑区域像素与非光斑区域像素的灰度差异平均提高了2.18倍,进而可利用阈值分割算法对光斑区域进行更准确的分割;最后,提出精度和召回率作为算法应用效果评价指标,将算法应用于1000张紫外图像帧,其平均值分别为0.9632和0.9827。由此可知算法实现了紫外检测图像中彩色光斑区域的准确分割,能够满足后续对紫外放电图像进行可靠诊断的要求,为基于紫外成像的电气设备放电诊断提供了可靠保证。

Ultraviolet(UV)imaging detection is an effective non-contact discharge detection method and the accurate segmentation of colored spots in UV discharge images is directly related to the accuracy of later diagnosis.Therefore,a method of color spot regions recognition and segmentation based on gray variance in UV image was proposed to accurately extract color UV spot regions.Firstly,UV discharge videos of electrical equipment were collected by UV imager,among which the color of the discharge spot area covers the colors commonly used in current UV imagers.Secondly,the processing method of UV video was improved.Gaussian function was used to pre-process the UV image,which increased the gray difference between the pixels of the spot region and the pixels of the non-spot region by an average of 2.18 times,and then the threshold segmentation algorithm could be used to segment the spot region more accurately.Finally,the precision and recall rate were used as the evaluation indexes of the algorithm application effect,and the algorithm was applied to 1000 UV image frames with the average values of 0.9632 and 0.9827 respectively.According to evaluation indexes,the algorithm achieved the accurate segmentation of colored spot regions in UV detection images,which could meet the requirements of the subsequent reliable diagnosis of UV discharge images and provide a reliable guarantee for the discharge diagnosis of electrical equipment based on UV imaging.