基于NSST图像融合的变电站开关状态识别

Switch State Recognition of Substation Based on NSST Image Fusion

针对变电站中开关状态图像识别易发生误判的问题,提出了一种基于非下采样剪切波变换(NSST)图像融合的变电站开关图像识别方法。该方法采用可见光和红外光双摄像模式,得到开关可见光图像和红外图像,根据两者的特点与互补特性,采用基于NSST的图像融合算法对可见光图像和红外图像进行融合,生成含有两种开关图像丰富细节信息和特征量的融合图像,采用改进加速稳健特征(SURF)算法对融合图像进行目标特征量提取和匹配,再采用基于混沌布谷鸟(CCS)算法的多阈值图像分割技术进行处理,最后基于霍夫变换得到开关臂和触点所在直线的斜率,根据两者角度差判别开关状态。仿真实验表明:所提方法对闭合与断开状态隔离开关的识别率分别为94.4%和100%,验证了所提方法的有效性。

In view of the problem of misjudgment of switch state image recognition in substation,a kind of method of switch image recognition of substation based on the non-subsampled shearing transform(NSST)image fusion is proposed.The method adopts visible and infrared dual camera mode to obtain visible and infrared image of the switch.According to the characteristics and complementary relationship of them,the visible image and infrared image are fused by using image fusion algorithm based on NSST and the fused image containing rich details and contour features of the two switch images is generated.The improved speeded up robust features(SURF)algorithm is used to extract and match the target characteristic quantity of the fusion image,and then the multi threshold image segmentation technology based on the chaotic cuckoo search(CCS)algorithm is used for processing.Finally,the slope of the line where the switch arm and two contacts are located is obtained on the basis of Hough transformand the switch state is judged in accordance with the angle difference between them.The simulation experiments show that the recognition rate of the proposed method for the disconnector at closing and opening state is 94.4%and 100%respectively and the effectiveness of the proposed method is verified.