Study On The Technology In Color Video Construction For Stationary Area In Night And Foggy Days

The promotion and application of clean energy such as natural gas,wind power and nuclear power put forward higher requirements on the level of safety precaution and control of their related facilities during operation.At present,most of the video surveillance systems that supervise the above facilities use a single infrared or visible band imaging technology,which cannot independently realize all-weather clear color video imaging of stationary area in night and foggy days due to their insurmountable limitations.Using the advanced multi-band colorization imaging technology,the same scene information collected by infrared and visible light imaging sensors of different wavebands is combined together by digital image processing technology and optical means to carry out the advantages of each other,and get a more informative video image,which can greatly enhance the performance of the video surveillance system performance and achieve good visual observation results.According to the requirements of the project,this paper designed and completed a set of color video imaging system in night and foggy days for camera fixed conditions with the characteristics of infrared imaging and visible light imaging.Combining the imaging advantages of infrared and visible light dual-band sensors,the system utilizes the front-end visible light sensor to carry out color video surveillance and extract part of visible light video sequences in the daytime with high visibility.Through the construction of mixed Gaussian background model,the scene information under the visible light imaging condition is obtained and used as the background of the final output color video sequence.Then,in the weather with insufficient visibility or night time,infrared sensor is used to extract the infrared video sequence,and the foreground object detection method combined with three-frame differential and Gaussian mixture model is used to extract the infrared object which needs to be observed.Finally,according to the registration parameters of the original infrared and visible video frame images,the infrared foreground target is merged into the visible scene proportionally based on the image mask operation to complete the construction of color video satisfying human visual requirements.Through the experimental analysis in different places and under different conditions,the feasibility and effectiveness of the system algorithm in this paper can be proved.The color video frame output by the system has a prominent target,a clear background and rich colors,and can fully reflect target and scene characteristic information without distortion,thereby providing a more realistic color visual effect.For a video sequence with a resolution of 720 X 576,the system algorithm in this paper can run at 40 frames per second,which can meet the real-time construction needs of color video.