The research process of intelligent monitoring image content analysis can be divided into four stages. One is to separate the target from the image. Objective classification of single target in simple and simple background environment;

The second is the analysis of the target state and behavior in the simple environment, such as the composition or dispersion of the direction, mode and goal of the movement; Then realize the behavior of detection and alarm abnormality; It is the main application in the current video monitoring field to achieve the target movement trajectory, realize the automatic tracking of the target and the statistics, correlation, filtering and trend prediction of the target behavior.

Three is in the complex environment, which is usually video monitoring environment, to achieve the above functions, (monophyletic, multi-source) of the image correlation, this is also a video monitoring image content analysis to practical key;

Four is video semantic interpretation, through the understanding of the image sequence, generate the image file of video semantic (correspond to image information), now in the big data become more mature, huge amounts of data is the basis of video semantic interpretation.

The first three stages are called video semi-semantic interpretation, mainly to determine whether there is "abnormal" situation in the image, and the image content cannot be understood sufficiently and accurately. The last kind of video semantic interpretation is the highest level of image content analysis, which is the foundation of big data, and there are many technologies to be solved.

At present, video monitoring image structure and description of content need visual explanation of help. Video monitoring intelligence is also a process of gradual development. It will not be achieved overnight, nor will it be the ultimate result. It will be through continuous accumulation and technological breakthrough. It needs to be pointed out that it is not true that video monitoring intelligence has been realized, and it is unrealistic to hope that it can solve all problems and deficiencies of the current monitoring system. At the same time, we should emphasize that intelligent application must be involved in the wisdom of human beings.