Identification of power poles based on satellite stereo images using graph-cut algorithm

摘要

Monitoring vegetation encroachment for high voltage transmission lines is a challenging job for electrical distribution companies. The encroachment interrupts electrical supply to industrial sectors and residential areas. It also endures a great cost to the authorities for maintenance and damage compensation. This brings a necessity for companies to monitor the vegetation near power lines. To solve these issues, existing monitoring methods are time consuming and expensive. In this paper, we proposed a new method to detect the vegetation using satellite stereo imagery. The method is based on Graph-Cut algorithm which applied on satellite stereo images. We incorporated the ordering constrain with Graph-Cut on stereo images to detect occlusion on stereo matching algorithm. This will increase the accuracy of the disparity map which is required task for identification of power poles near vegetation based on satellite stereo images. In conclusion, based on the results, Our proposed Graph-Cut algorithm produces more accurate disparity maps than the existing matching algorithm used for disparity calculation.