Study Of Ballistics Terminal Target Gesture Recognition Technology

Based on the pose recognition of fighter through using the measuring information of infrared imaging seeker during partial image tracking, the aim point parameters can be estimated more precisely and the accurate attack for vulnerable part of target can be realized. The pose recognition technology of fighter was studied by using infrared imaging seeker information on the background of Guidance Integrated Fuzing technology.On the analysis of the characters of infrared image sequence of seeker during high-speed encounter, noise reduction technology, image sequence segmentation technology and the blob eliminating technology etc were used to preprocess the image sequence of target. Then the improved Otsu algorithm was simulated and verified.Target recognition can be achieved by BP (Error Back Propagation) neural network. Aim at fighter which has been translated, rotated, scaled or affine changed, Hu and Affine invariant moments were used to describe the shape feature of target, then fighter can be recognized by using BP neural network in order to get the length parameter of airplane-axis.This paper extracts skeleton of a target image by a thinning algorithm and detects airplane-axis included in images by the Hough transform. The coordinate parameters of airplane nose and tail can be obtained through the position of airplane nose and tail included in the image and measurement information of the imaging seeker. Fighter nose, tail and the center of seeker focal plane can definite a series of plane during encounter, the intersecting line of which is fighter axis, through the identification of fighter axis direction vector in the space, pose of airplane can be recoginzed.Delay time can be revised according to the pose recognition, the length parameter of fighter axis and the relative velocity vector of missile and target.The image preprocessing algorithm, target recognition algorithm, skeleton extraction algorithm and fighter axis detecting algorithm were simulated and verified. Then interfaces of image preprocessing algorithm and fighter axis detecting algorithm were designed in C++Builder 5 platform. The effects because of threshold selection error and different encounter conditions for pose recognition and delay time correction were discussed.