风电叶片图像直线特征检测与拼接方法  

Straight Line Features Detection and Mosaic of Wind Power Blades Image

作　　者：马宝琰 汤磊 赵晶[1] 何勇军[1] MA Bao-yan;TANG Lei;ZHAO Jing;HE Yong-jun(School of Computer Science and Technology,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学计算机科学与技术学院,哈尔滨150080

出　　处：《哈尔滨理工大学学报》2020年第5期83-92,共10页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61673142)。

摘　　要：风电机组运行过程中需要定期巡检叶片表面。人工巡检方式工作量大、成本高且检出率低,迫切需要引入智能巡检系统。风电叶片图像的拼接是智能巡检系统的关键环节,直接决定着系统性能。目前的拼接方法在面对风电叶片这样结构与颜色单一的条形对象时无法提取出足够的特征点,拼接成功率低。针对这一问题提出了基于直线特征的拼接方法。该方法首先检测风电叶片图像中的直线,然后筛选去重作为特征进行图像配准,最后根据平移旋转矩阵拼接叶片图像。实验结果表明,所提出的方法能适应拼接结构单一、重合度小和背景多样等复杂的情况,具有较强的鲁棒性和稳定性。It is necessary to periodically inspect the blade surface of the wind turbines operation. The widely used manual inspection method brings a heavy workload, high cost and low detection rate, so it is urgent to need anautomatic inspection system.Image mosaic of wind power blades is a key part of the intelligent inspection system, which directly determines the system performance. Current splicing methods cannot extract enough feature points when facing a strip object with a single structure and color such as wind power blade, causing failure in splicing success. A splicing method based on line features is proposed to solve this problem. This method first detects straight lines in the images of the wind power blade, then selects the de-duplication as the feature for image registration, and finally splices the blade image according to translation rotation matrix.Experimental results show that the proposed method shows strong robustness and stability in complex situations such as single structure, small coincidence degree and diverse backgroundstability.

关 键 词：图像拼接 直线特征 差异度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]