Image Processing Based On Smart Grid

In order to meet the needs of social development, Smart Grid integrates the traditional physical grid with advanced monitoring technology, information technology and computer technology, forming a new type of highly efficient power grid. In the situation of high voltage power equipment running throughout the year, with the aging foundry’s equipment and insulation damage, the requirements for preventive maintenance becomes increasingly demanding. Due to the expanding size of the transmission lines and facilities that are hard for artificial monitoring, using computers and intelligent devices to automatically obtain and analyze the monitoring pictures emerges as time requires. This thesis quickly determines whether there is corona and arc discharge area in the UV image by analyzing the UV image, and in order to identity the insulation parts, labels the discharge area through image automatic annotation. This work can not only help people find the safety risks in early time, but also provide the failure messages to facilitate repairing, which can avoid catastrophic accidents. The major work of this thesis is as below:1. Traditional image automatic annotation labels the image pixels, and in order to reduce the computation complexity, this thesis segments the image into superpixels, which is composed of some pixels, which have the same image feature, preparing for the image automatic annotation for Smart grid image.2. In purpose of analyzing the UV image automatically and providing early warning based on the recognition result, this thesis builds a process to detect the corona and arc discharge area in the UV image, and gets preferable recognition result through it.3. This thesis labels the UV image to identify the discharge area. In order to reduce the computation complexity, this thesis combines the Error Correcting Output Code with image automatic annotation, turning the multi-classification into binary classification, and labels the superpixels which have been segmented in the previous step.