Moving target detection in infrared video _ IR Video

The application background of infrared video moving target detection: the usual video target detection, often is the visible light target detection, this detection and tracking technology has been very mature, but there is an unavoidable flaw, it is not in the light of the situation to carry out effective detection, in the night without light can not be detected at all. and infrared detection can make up for such deficiencies. Infrared video is imaged by the intensity of the infrared radiation of the object, thus escaping the shackles of visible light, which can be used to detect and trace all day.
Some features of infrared video detection: Infrared video target detection, compared to the target detection in the visible light is much more difficult, because: 1, the quality of infrared video is very easy to take the infrared video of the impact of equipment, once the performance of the device is not stable, the infrared video will be taken out in the back of the detection         Process is very difficult, so the infrared video camera equipment than the visible light requirements higher. 2, infrared video imaging relies on the infrared radiation intensity of the object to image. In principle, when the temperature is greater than absolute 0 degrees will radiate infrared rays, the higher the temperature, the stronger the radiation, the more clear imaging. Therefore, in the infrared imaging process, if the target temperature and background very close to the case, detection will bring difficulty. But under normal circumstances, if the moving target is a non-life body, then it has power source, and the power source temperature must be higher than the ambient temperature, which increases the probability of successful detection.         If the moving target is a living body, because the life body has its own stable temperature, for example, the average temperature of people is about 36.5 degrees, animals such as pigs is about 39 degrees, so in the imaging of each frame of the image, the target pixel value is maintained, if the life of the gray value is changing, it is not a life-body target. 3, in the infrared video of the moving target, in the process of movement, the radiation intensity and background of the radiation intensity difference is changing, the reason is that the radiation intensity of the target is unchanged, and the background of the radiation intensity due to changes in the environment, such as the original side without street lights, when the car to the place where the lights, the background has changed, This shows in the video that the target can be easily detected, sometimes because it is too similar to the background to detect.
Some suggestions on infrared video target detection: 1, optical flow detection algorithm: The optical flow detection algorithm itself is not targeted at the visual or infrared video, the principle is still general.          In the process of optical flow detection, the use of corner detection, calculation of optical flow, the use of moving speed to determine whether the target or background, and then detect the target.          2, pixel value judgment: Because the target's infrared radiation intensity and background is definitely different, therefore, judge a moving target in the video pixel value range, can effectively avoid the target error detection, but this method is based on the premise that can find a suitable range of pixel values.          3, Edge judgment (is also the size of the judge): Detection of the target is the edge, if the detection of the target is not the edge, or the edge is particularly large, beyond the threshold set by themselves, then it is not the goal. 4, if the background is moving, then you can also consider the angle of judgment. As for the angle how to set, according to the specific video, according to the motion of the video, is the rotation or translation, its angle judgment is different.