Method for deleting video big data redundancy based on information divergence

The invention discloses a method for deleting video big data redundancy based on information divergence, and is aimed at providing a redundant video data deleting method, saving storage capacity, reducing the analysis time of video big data, improving the efficiency of maintenance and management of video big data. According to the technical scheme, first video big data are expressed with a video big data set V; the V is normalized; then the K-L divergence for a video image frame vector and an information quantity distribution matrix M_L are defined; the contribution of each video image frame vector to the information quantity of the V is confirmed; a redundant video image frame making the smallest contribution to the information quantity is deleted from the V. The method can alleviate the storage pressure of video data, improve the storage efficiency of video data, speed up the calculation and analysis rate of video big data, and improve the operational efficiency of the whole big data video application syste