Research On Content Based Video Indexing And Application

Helicopter inspection and drone inspection have become rather important methods of power grid inspection,playing a critical role maintaining the safety and stability of power grid system.However,processing the raw footage is still mostly artificial,main disadvantages including overwhelming workload,high undetected rate and low efficiency;a video indexing system would significantly increase efficiency and lower undetected rate.This article has implemented video indexing to aerial photography based on its content,achieving video searching for power grid aerial surveillance videos.This article is setting up video indexing including towers and insulators applying a searching method based on content,achieving auto-searching of aerial photography.Based on the distinguishing features of surveillance videos,we could take out the key frames using a video segmentation method based on Fourier Descriptor.Secondly,apply LSD Line detection,Harris Corner Detection Algorithm,and Convolutional Neural Network identifying the shots of towers and insulators,and set up video indices of towers and insulators.Then set up an index and searching system of power grid inspection videos according to the characteristics of such videos.The system would mainly consist of video index module,video searching module and system database.Finally,apply a power grid inspection video as an example to test the system,outcomes have suggested such system being effective and practical.