基于视觉显著性的飞机蒙皮缺陷检测方法  

Aircraft skin defect detection method based on visual saliency

作　　者：张研 田军委[1] 刘青[2] 贾昆 刘雪松 ZHANG Yan;TIAN Junwei;LIU Qing;JIA Kun;LIU Xuesong(School of Defence Science and Technology,Xi'an Technological University,Xi'an 710021,China;School of Automation and Information Engineering,Xian University of Technology,Xi'an 710054,China;Inner Mongolia North Heavy Industry Group Co.LTD,Baotou 014033,China)

机构地区：[1]西安工业大学兵器科学与技术学院,西安710021 [2]西安理工大学自动化与信息工程学院,西安710054 [3]内蒙古北方重工业集团有限公司,包头014033

出　　处：《激光杂志》2022年第3期61-64,共4页Laser Journal

基　　金：国家自然科学基金项目(No.51875447);陕西省重点研发计划(No.S2018-YF-ZDGY-0583)。

摘　　要：飞机蒙皮缺陷是导致航空安全事故的重要因素之一。针对当前蒙皮缺陷依赖人工目视检测,存在效率低、实时性差、人工经验敏感等问题,提出一种基于局部对比度的显著性模型缺陷检测方法。算法采用自动γ校正算法对输入缺陷图像进行修正,增强缺陷与背景的动态范围;在基于局部对比度的显著性检测模型中融入HSI颜色空间中的I分量以提升缺陷的显著性检测效果,结合自适应阈值的Canny边缘检测算子检测出显著图的缺陷边缘。实验结果证明,提出的显著性算法与传统检测算法相比,能更为清晰地表现出缺陷特征,且对光源分布不均匀的缺陷图像的检测结果也有较好的鲁棒性,对In与Sc两种缺陷的检测成功率分别为80.7%和92.3%,高于传统算法,为飞机蒙皮智能缺陷检测提供了良好的依据。Aircraft skin defect is one of the important factors leading to aviation safety accidents. In order to solve the problems of low efficiency, poor real-time performance and sensitivity of human experience, a significance model defects detection method based on local contrast was proposed. The automatic gamma correction algorithm was used to correct the input defect image and enhance the dynamic range of the defect and background. The I component of HSI color space was incorporated into the saliency detection model based on local contrast to improve the saliency detection effect of defects, and the defect edges of saliency graph were detected by Canny edge detection operator combined with adaptive threshold. Experimental results show that compared with the traditional detection algorithm, the significance algorithm proposed to this paper can show the defect characteristics more clearly, and has good robustness to the detection results of the defect images of uneven light source distribution. The detection success rate of In and Sc defects is 80.7% and 92.3% respectively, higher than the traditional algorithm. It provides a good basis of aircraft skin intelligent defect detection.

关 键 词：显著性检测 自动γ校正 局部对比度 CANNY算子 

分 类 号：TN209[电子电信—物理电子学]