Vision-based Moving Target Detection And Location Technology

This Vision-based moving target detection and locating technology is widely used in civil transportation,intelligent driving,anti-disaster and other tasks.In military,unmanned vehicles with photoelectric sensing platform has a unique advantage,especially in the real-time,zero-injury,all-weather effective monitoring and control of movable,masked,time-sensitive ground targets or surface targets.The monocular vision system is installed on an amphibious micro-unmanned vehicle,performs a reconnaissance system that monitors the ground and the water surface,detects moving targets from the captured video images,automatically adjusts the car-mounted pan-tilt to track moving targets,and simultaneously positions the moving targets in three dimensions.Use location information to guide the control of amphibious micro-unmanned vehicles to implement further solutions.The thesis discussed the problem of the detection and location of moving targets in dynamic background,The main contents list as follows::Visual system design and image processing: Research on vision system design and image processing technology is prepared for follow-up work.The visual perception system designed uses CMOS sensor + laser range finder + ARM processing chip + display terminal(computer)solution.Image processing technology involves image graying,binarization,image denoising,edge detection and image defogging.At the same time,the experimental results of the corresponding algorithms are given.Target detection: In order to realize moving target detection under dynamic background,a moving target detection algorithm based on single Gaussian background model and continuous frame difference method is proposed.The experimental results based on improved algorithm are given and used only.The experimental results of the difference scene algorithm based on the single Gaussian background model are compared.The trajectory pre-pushing aspect: the dual-element pyroelectric infrared sensor is equipped with an infrared lens to form a "meter"-shaped sensing structure composed of detector elements,and then the layout of the perceptual model is studied to determine the square of the four amphibious micro-unmanned vehicle layouts.The perceptual model can meet the requirements of the measurement of the moving target trajectory,and the moving target trajectory pre-rendering algorithm under the model is explained in detail.The effectiveness and enforceability of the algorithm are verified by outdoor experiments.Target positioning: Firstly,the key information needed to achieve positioning is clarified.On the basis of this,in order to recover the missing target depth information in the 2D image,a ranging algorithm based on monocular ranging model and a complete extraction contour are proposed.Target ranging algorithm.According to the experimental results,more accurate distance information can be obtained by using the designed monocular ranging algorithm.