CN111159470A - Monitoring video content filtering method - Google Patents

Abstract

The invention provides a monitoring video content filtering method, which realizes monitoring of key areas by performing 4 aspects of element analysis on monitoring videos, including analysis on whether a person, an object or a vehicle enters or leaves a preset area, whether the person stays in the appointed area for a long time and wanders, whether video images have great changes and whether a group aggregation phenomenon exists. Sending out alarm signals when the conditions meet 1 or more than 1 element; and directly neglecting the videos which do not conform to the characteristics, thereby improving the processing efficiency of the monitoring videos.

Description

Monitoring video content filtering method

Technical Field

The invention relates to the technical field of video monitoring of security systems, in particular to a method for filtering monitoring videos, which omits useless information so as to save management cost.

Background

Technologies such as high-definition video, video storage, intelligent video analysis and the like become main aspects of the current video technology development. With the popularization of a large number of video monitors, how to realize the quick browsing of massive videos, the concentrated summarization of massive videos and the quick retrieval of required video data in massive video data become the most important research contents in the current video field, and especially the important research contents are in traffic and security video applications.

Disclosure of Invention

The invention aims to provide a monitoring video content filtering method based on traditional video monitoring, which omits a large amount of information which is useless for security protection in a monitoring system by intelligently analyzing videos, thereby saving a large amount of manpower and material resources when managing and using the videos.

The technical scheme of the invention is as follows:

a monitoring video content filtering method is characterized in that: monitoring of key areas is realized by performing 4 aspects of element analysis on the monitoring video; the 4 aspects of element analysis comprise:

(1) analyzing whether a person, an object or a vehicle enters or leaves a predetermined area;

(2) whether a person lingers in a designated area for a long time;

(3) whether the video image has great change;

(4) whether a population aggregation phenomenon exists;

sending out alarm signals when the conditions meet 1 or more than 1 element; and directly neglecting the videos which do not conform to the characteristics, thereby improving the processing efficiency of the monitoring videos.

According to the invention, whether a person, an object or a vehicle invades or crosses a border in the video or not is monitored, whether the person stays in a designated area for a long time and wanders, whether the video image has great change or not and whether a group aggregation phenomenon exists or not are monitored, abnormal behaviors in the video are analyzed, the perimeter precaution of a key area is realized, a large amount of information which is useless for security in a monitoring system is omitted, and thus a large amount of manpower and material resources are saved.

Detailed Description

The invention realizes the monitoring of key areas by performing 4 aspects of element analysis on the monitoring video; the 4 aspects of element analysis comprise:

(1) it is analyzed whether a person, object or vehicle enters or leaves a predetermined area.

Whether a person, an object or a vehicle enters or leaves a predetermined area is analyzed, and the method is realized by the following steps:

(11) tracking a moving target in the video, judging whether the moving target belongs to a monitoring area or outside the monitoring area, and recording an external rectangle of the moving target in real time;

(12) real-time detection is carried out in the tracking process, if the moving target belongs to a moving target outside a monitoring area, the vertex of an external rectangular frame of the moving target is simultaneously positioned in the area and outside the area in the moving process, and at the moment, the fact that outside-in invasion occurs can be judged;

(13) and carrying out real-time detection in the tracking process, if the target belongs to a moving target in the monitoring area, and the vertexes of the external rectangular frame of the target meet the requirement that the target is positioned outside the area and in the area in the moving process, at the moment, judging that the target is invaded from inside to outside.

(2) Whether a person lingers in a designated area for a long time;

the specific method for analyzing whether a person stays in the designated area for a long time or not is as follows:

(21) the main direction angle difference of n1(n1 is more than or equal to 2) curves on the target motion track is more than 120 degrees;

(22) in the whole moving process of the target, the moving distance d of a curve with N2(N2 ═ 3) in an N2(N2 ═ 15) frame is stable in a range interval [ d1, d2 ];

(23) in the whole moving process of the target, the distance between the position of the current frame of N3(N3 is 3) times and the initial position of the monitored area is smaller than that of the previous N3(N3 is 10) frame;

(24) and determining the target simultaneously meeting the characteristics as the abnormal behavior of detention wandering.

(3) Whether there is a large change in the video image.

Whether the video image has great change or not is analyzed from two aspects of shielding and great movement of a camera, and the specific method comprises the following steps:

(31) camera covering situation: if the sudden loss of the picture is detected, but the statistical variance of each pixel value of the picture is not 0, and the picture change difference in the subsequent t frames is smaller than a given threshold value;

(31) camera steering situation: and judging by combining the background difference degree and Surf feature point matching, namely when the proportion of the difference pixel points of the current image and the average image in a period of time is greater than a given threshold value, using Surf feature matching, and judging that the direction is turned if the positions corresponding to a plurality of matched feature points in the matching result are overlapped.

(4) Whether there is a population aggregation phenomenon.

Whether a population aggregation phenomenon exists is analyzed, and the specific method is as follows:

(41) when the crowd aggregation is not generated, the crowd is dispersed and ordered, if the crowd is dense, the mask communication area expands to a certain degree, and at the moment, the crowd aggregation is judged when the proportion of the area of the mask communication area and the area of the area occupying the background area reaches a given threshold value;

(42) tracking multiple targets in a video scene, and if the difference between the number of the targets entering the video scene and the number of the targets disappearing is larger than a given threshold value, judging that the crowd gathers.

When the monitoring video is intelligently analyzed, an alarm signal is sent out under the condition that the monitoring video meets 1 or more than 1 element; and the video which does not conform to the characteristics is directly ignored, so that the processing efficiency of the monitoring video is improved, and a large amount of manpower and material resources are saved.

Claims (5)

1. A monitoring video content filtering method is characterized in that: monitoring of key areas is realized by performing 4 aspects of element analysis on the monitoring video; the 4 aspects of element analysis comprise:

(1) analyzing whether a person, an object or a vehicle enters or leaves a predetermined area;

(2) whether a person lingers in a designated area for a long time;

(3) whether the video image has great change;

(4) whether a population aggregation phenomenon exists;

sending out alarm signals when the conditions meet 1 or more than 1 element; and directly neglecting the videos which do not conform to the characteristics, thereby improving the processing efficiency of the monitoring videos.

2. The surveillance video content filtering method according to claim 1, characterized by: whether a person, an object or a vehicle enters or leaves a predetermined area is analyzed, and the method is realized by the following steps:

(11) tracking a moving target in the video, judging whether the moving target belongs to a monitoring area or outside the monitoring area, and recording an external rectangle of the moving target in real time;

(12) real-time detection is carried out in the tracking process, if the moving target belongs to a moving target outside a monitoring area, the vertex of an external rectangular frame of the moving target is simultaneously positioned in the area and outside the area in the moving process, and at the moment, the fact that outside-in invasion occurs can be judged;

(13) and carrying out real-time detection in the tracking process, if the target belongs to a moving target in the monitoring area, and the vertexes of the external rectangular frame of the target meet the requirement that the target is positioned outside the area and in the area in the moving process, at the moment, judging that the target is invaded from inside to outside.

3. The surveillance video content filtering method according to claim 1, characterized by: the specific method for analyzing whether a person stays in the designated area for a long time or not is as follows:

(21) the main direction angle difference of n1(n1 is more than or equal to 2) curves on the target motion track is more than 120 degrees;

(22) in the whole moving process of the target, the moving distance d of a curve with N2(N2 ═ 3) in an N2(N2 ═ 15) frame is stable in a range interval [ d1, d2 ];

(23) in the whole moving process of the target, the distance between the position of the current frame of N3(N3 is 3) times and the initial position of the monitored area is smaller than that of the previous N3(N3 is 10) frame;

(24) and determining the target simultaneously meeting the characteristics as the abnormal behavior of detention wandering.

4. The surveillance video content filtering method according to claim 1, characterized by: whether the video image has great change or not is analyzed from two aspects of shielding and great movement of a camera, and the specific method comprises the following steps:

(31) camera covering situation: if the sudden loss of the picture is detected, but the statistical variance of each pixel value of the picture is not 0, and the picture change difference in the subsequent t frames is smaller than a given threshold value;

(31) camera steering situation: and judging by combining the background difference degree and Surf feature point matching, namely when the proportion of the difference pixel points of the current image and the average image in a period of time is greater than a given threshold value, using Surf feature matching, and judging that the direction is turned if the positions corresponding to a plurality of matched feature points in the matching result are overlapped.

5. The surveillance video content filtering method according to claim 1, characterized by: whether a population aggregation phenomenon exists is analyzed, and the specific method is as follows:

(41) when the crowd aggregation is not generated, the crowd is dispersed and ordered, if the crowd is dense, the mask communication area expands to a certain degree, and at the moment, the crowd aggregation is judged when the proportion of the area of the mask communication area and the area of the area occupying the background area reaches a given threshold value;

(42) tracking multiple targets in a video scene, and if the difference between the number of the targets entering the video scene and the number of the targets disappearing is larger than a given threshold value, judging that the crowd gathers.