Device based on video smoke detection, and method based on video smoke detection

Abstract

The invention discloses a device based on video smoke detection, and a method based on video smoke detection. The method includes the following steps: scattered foreground points are firstly acquired through a video image acquisition unit and a foreground acquisition unit; de-noising processing of the foreground points is then carried out through a de-noising unit; the foreground points are then accumulated through a foreground accumulation unit; a region connection unit is utilized to connect the accumulated foregrounds and acquire relevant parameters; the average gradient information of connected regions is acquired through a gradient computing unit; the relevant parameters are then set through a false smoke removal unit to remove some false smoke regions; feature matching of suspected smoke regions that are not removed is then carried out through a feature matching unit; a judging and alarming unit finally decides whether an alarm is to be sent by judging whether the matching times exceed a threshold value or not. The invention has the advantages that the foregrounds are acquired through the three frame difference method and accumulated through a fixed window, false smoke is then removed through the multi-feature fusion method, and finally the matching times are judged to decide whether the alarm is to be sent or not.

Description

[0001] technical field [0002] The invention relates to the field of computer vision, in particular to early smoke detection applied in the field of video surveillance. [0003] Background technique [0004] Fire prevention and detection has always been the focus of people's attention. When a fire occurs, the earliest phenomenon is smoke. Therefore, the detection of smoke is more important. Most of the detection methods currently used are divided into: smoke detectors and video surveillance. [0005] All kinds of smoke detectors require smoke to reach a certain concentration and reach a certain range of the detector before they can give an effective alarm. When facing tall and open places, this kind of detector is not effective enough. With the rapid development of modern society, tall and large buildings are increasing day by day. Once a fire breaks out, when this kind of detector sends out an alarm, great losses have been caused, and the function and purpose of the o...

AI Technical Summary

Problems solved by technology

Once a fire breaks out, when this kind of detector sends out an alarm, it has caused great losses and lost the original detection function and purpose.

[0006] For ordinary video surveillance, a large number of personnel are required to carry out all-round and all-weather real-time monitoring

This approach is more difficult to implement

One is that a large number of personnel are required to observe many videos in real time; the other is that it is impossible for monitoring personnel to concentrate on the whole day

Images