Target Location And Identification Based On Video Stream

Target location and identification are widely used in various fields.Because of the diversity of target types and complexity of background,different target location and recognition algorithms are designed for different applications.With the information of target movement,target location and identification determine the position and identity of target from video stream by computer vision and pattern recognition algorithm.Such technology is widely used in virtual reality,mobile robot tracking and other fields.Therefore,it is worthwhile to study the target location and identification based on video stream.In this thesis,target location and identification system including hardware architecture and corresponding software algorithm based on video stream is designed and implemented for warfare training system.During warfare training,image acquisition module monitors the battlefield real-timely.Soldiers appeared in sequence of image is discovered and their position is located.Their identification code is read out.Affine transform is applied to calculate soldiers' position in battlefield from video stream.Near-Infrared LED is used as light source for target identification.It emits light pattern according to pre-determined identity code.Iterative thresholding method is used to preprocess the image and segment the light spot.The algorithm of track correlation detection is implemented to detect the position of the target in the image.Least squares method is used to predict the target movement and remove false target spots.Target trajectory is established and its light pattern is saved.Identity code of every target can be read out from light pattern information.In order to reduce error read out rate,parity check is used in the paper.In order to improve the accuracy of coordinate mapping,nonlinear image distortion is corrected by template correction algorithm.A block iterative correction algorithm is proposed to further reduce mapping error.In the algorithm,image is evenly segmented into blocks firstly;then,iterative parameter calculation is used based on constraints between image blocks and statistical principle.The algorithm reduces effect of artificial error and system error,and accurately map the coordinates from the distorted image to the real scene.