基于聚类分析和骨架提取的含裂纹超声红外热像增强方法研究北大核心CSCD

超声红外热图像因噪声干扰及缺陷位置的热扩散,导致其存在对比度差、清晰度低、边缘模糊等问题。为了增强红外图像视觉效果,提高缺陷检测能力,提出了一种基于聚类分析和缺陷骨架的超声红外图像增强方法。采用基于kmeans的DBSCAN聚类算法对裂纹发热区域进行识别聚类,将图像分解为缺陷生热区域与非缺陷区域;然后,对缺陷区域进行骨架描述,并沿裂纹骨架走向采用改进的部分子块重叠直方图均衡算法对缺陷图像进行增强。提出的超声红外图像增强方法与常用的直方均衡化、限制对比度自适应直方图均衡化、自适应同态滤波三种方法进行对比,结果表明所提的增强方法可以得到对比度更显著的图像,具有明显的优势。提出的方法为增强超声红外图像视觉效果、提升裂纹诊断能力提供了一种有效方法。

Crack enhancement method for ultrasonic infrared thermal images based on cluster analysis and skeleton extraction

Due to thermal diffusion at defects and noise interference,ultrasonic infrared thermal images are subject to poor contrast,low definition and blurred edges.In order to enhance the visual effect of infrared images and improve the ability detecting defects,a method based on cluster analysis and defect skeleton is proposed to enhance ultrasonic infrared images.The DBSCAN clustering algorithm based on k means is employed to identify and cluster the hot spot of the crack,so as to decompose the image into defect area generating heat and non defect areas.Then,the direction of the crack skeleton is described by skeleton extraction.Along the skeleton direction,the defect image is enhanced with an improved partially overlapped sub block histogram equalization algorithm.Finally,the proposed method is compared with commonly used algorithms,like histogram equalization,contrast limited adaptive histogram equalization,and adaptive homomorphic filtering.The results show that the proposed enhancement method can obtain images with more significant contrast and has obvious advantages.The proposed method provides an effective method for enhancing the visual effect of ultrasonic infrared images and improving the ability of crack diagnosis.