Intelligent video analysis method

Abstract

The invention proposes an intelligent video analysis method. The method comprises the following steps of S1, creating a training sample according to an obtained video image; S2, extracting a characteristic according to the training sample; S3, establishing a training classification model according to the extracted characteristic; and S4, detecting a target according to the training classification model. The method has a capability of intelligently analyzing, acquiring, storing and transmitting large-scale diversified basic videos, an image big data management capability, and integrative capabilities of interconnection with other business systems and the like. Characteristic description information can be stored to form an index, and a video of a concerned event can be quickly located and played from massive video files by effectively utilizing and managing stored video file resources, so that the video record reverse query efficiency is improved and intelligent management of video data is realized.

Description

technical field [0001] The invention relates to the technical field of video analysis, in particular to an intelligent video analysis method. Background technique [0002] With the advancement of my country's informatization process, it is undeniable that the construction of a large number of front-end high-definition video equipment has greatly improved the quality of the entire video surveillance. Meet the viewing needs of front-end video for personnel at all levels. A large number of regulatory work often faces various emergencies. In actual operation, due to the relatively backward technology of the video surveillance application system and its independence from the business system, the implementation effect of the crisis management mechanism is greatly restricted. [0003] The current video surveillance system mainly has the following problems: [0004] 1. There are many real-time monitoring screens, all rely on manual identification. [0005] 2. Massive video recordi...

AI Technical Summary

Problems solved by technology

[0004] 1. There are many real-time monitoring screens, all relying on manual identification

[0005] 2. Massive video recordings, unable to quickly locate

[0006] 3. The integration efficiency of video surveillance resources is low, and a unified resource pool cannot be formed

[0007] 4. Video surveillance and business systems are relatively independent and cannot be linked

[0008] 5. Video surveillance is only used for time-based evidence collection, and early warning is difficult

[0009] 6. The image resources are not converted into information materials, and the actual combat assistance is not enough

Images