Research On Infrared Image Background Modeling And Suppression For Gas Leak Detection

In our highly industrialized modern society,we used more and more kinds of gas,most of these gases are poisonous or explosive.If these gases get leaked,it will cause serious problems to the society.In this paper,based on this,we studied an infrared i mage detection algorithm combining background modeling and background suppressi on.First of all,we study the characteristics of the infrared gas leak image,there is no characteristic information such as color,texture,it has no profile,no liquidity rules,that makes it too difficult to improve the gas leak detection.Therefore,the traditional target detection method can not meet the detection requirements of the infrared image of the leakage gas.In the practical application,not only we should realize the accuracy of the test results,and also the real-time performance of the system.Therefore,the detection of leakage gas is a challenging problem.Then,in view of these characteristics of the infrared video of leaked gas image.This paper selected the passive gas detection technology.And based on the study of comparing a variety of background modeling algorithm.Due to the excellent features of VIBE algorithm,.VIBE has been deeply analyzed and studied in this paper.On this basis,the traditional VIBE algorithm is improved,an infrared image background modeling algorithm based on fast foreground model is proposed.We apply fast foreground modeling method to the background modeling.We significantly reduced the comparison times of foreground points in the VIBE algorithm,and the experimental results showed that the improved VIBE algorithm could greatly improve the detection speed.At the same time,this paper also studys an automatic threshold calculation method to infrared gas leak detection algorithm,by replacing the fixed threshold value in traditional VIBE algorithm with the method of automatic computing threshold,the adaptability and accuracy of this algorithm are improved.The experiment results shows that,compared with the classical VIBE algorithm,the improved VIBE algorithm we proposed,can not only achieve better real-time performance,but also has higher accuracy and robustness,and can satisfy therequirement of a variety of scenarios of infrared gas leak image detection.Finally,in order to remove the noises and the fake target areas in the target image we get after background modeling.In this paper,we studied and compared the traditional background suppression methods,and an improved morphological filtering algorithm are studied,the experiments show that the proposed improved morphology filtering algorithm has better accuracy and can get a real time performance,and meet the requirements of infrared gas leak image detection to background suppression algorithm.