Recognition Method of Insulator Based on Object Proposals and Structure Research
Aiming at the low accuracy of the insulators recognition in aerial image and the limited speed of using traditional sliding window algorithm,a step by step insulator recognition method based on object proposals algorithm and structure research was proposed.First of all,reduce the amount of detection windows by using the object proposals algorithm based on the gradient profile closed.Second,extract the local feature to train the classifier and recognize the detection window for achieving insulator' s coarse positioning.Finally,the insulator skeleton structure feature was calculated,which was used to search the insulator more specifically.In the experiment,the insulator recognition rate reached 89％,and was promoted 9 times compared to the traditional sliding window method.The results show that the proposed algorithm has good recognition accuracy and rapidity.
insulator recognitionobject proposalsmachine learningstructure research
NETL
NSTL
万方数据
维普
