结合角点的边缘检测法实现输电线路杆塔标牌图像定位

Edge Detection Method Combining Corner Points to Realize the Image Location of the Tower Sign in the Transmission Line

针对现有的定位算法鲁棒性较低,泛性不佳等问题,以手持设备获取的杆塔标牌图像为研究对象,提出了改进边缘检测法。通过透视变换对图像进行倾斜矫正处理,根据像素累计情况采用边缘检测法实现标牌区域的初步定位,结合Radon角点检测结果,计算最小欧几里得距离进一步剔除背景干扰区域,使得标牌区域定位结果更加精确,并通过光学字符识别(OCR)加以验证。试验结果证明,该方法具有较高的准确性和较强的工程价值。

Aiming at the problems of low robustness and poor generality of the existing localization algorithms,this paper takes the image of pole tower sign obtained by handheld devices as the research object,and puts forward the improved edge detection method.Through the perspective transformation to the image tilt correction processing,according to the pixel accumulation using edge detection method to achieve the initial location of the sign area,combined with the Radon corner detection results,calculate the minimum Euclidean distance to further eliminate the background interference area,so that the sign area location results more accurate,and through OCR character recognition to verify.Tests show that this method has high accuracy and strong engineering value.