Image Recognition And Processing Technology In Substation Intelligent Inspection System

Greater demands are being placed on the reliability and safety of all aspects owing to Smart Grid.Inspection is one of the most important ways to ensure the safety and stability of key equipments in the substation.Intelligent inspection can improve a lot in inspection efficiency and real-time operation than traditional manual ways,and it saves more operating costs.The methods of image recognition is pivotal element in intelligent inspection system.The recognition methods of dial instrument,digitron and protective strap in substation are implemented,and the methods are described in this thesis.The dial instrument is located by matching the feature points in target image with those of template image.Pointer skeleton can be get by thinning the binarized image,and the reading of the instrument will be calculated by angle of pointer skeleton and labeled pointes in final.The screen is located by matching with template image,reading numbers will be recognized by Three-line Scanning Method with binarized image.The image of protective strap is binarized by the pre-set color and thresholds,the edge is detected by morphological processed connected area,the status of each strap is judged by the angle of edge's minimum area external bounding rectangle.Image recognition method's accuracy of dial instrument reached the acceptance criteria of Power Corporation according to the sample images and their distorted derivation.Image recognition method of digitron is widely used in other fields,but it performs not well in our poor samples,the result is lack of stringency.The image recognition method of protective strap performs well in samples which obtain only one type,the method is highly dependent on the type of strap,it should be optimized to improve the degree of generalization.