A Study On Military Objects Recognition And Tracking Based On The Gabor Filter

This dissertation is an exploration on the Gabor filter used in military objects recognition and tracking . Through the analysis of the Gabor function, the properties of the Gabor filter, such as the invariation of scale, rotation and translation, were validated from both aspects of the theory and the experiments. Based on the analysis of the response values that were output from the Gabor filter, the local maximum values were found by using an adaptive window. And these local maximum points were taken as the feature points of the object. By matching the sum of the energy values in a small neighboring area around every feature points, the object recognition was realized. At the same time, the reconstructed object using the feature points was taken as a tracking model, and the real-time tracking was realized by use of the adaptive deformable model.This thesis mainly consists of the following parts:(1) Through the study of Gabor function, make a understanding of the structure and parameters of the Gabor filter whose mother function is an odd-Gabor wavelet. The properties of scale, rotaion and translation invariation of the filter were validated by the experiments.(2) The local maximum points in different orientations and scales were extracted from the fitered image, and getting a feature point by summing the energy values in a small neighboring area around each feature point. The object recognition was realized by a hierarchical matching structure in which a low frequency matching was carried on firstly and then a high frequency matching was put up in succession.(3) Selecting the feature points output by the Gabor filter with an auto-adaptive searching window, and reconstructed the object image with these feature points.Taking the reconstructed image as a tracking model, in which the parameter of scale and orientation can be changed during the tracking process, then the robusticity of the tracking algorithm was improved.The innovations of this thesis include the following parts:(1) An adaptive searching algorithm was proposed for correctly searching the local maximum response value of the Gabor filter, and its feasibility was validated with the experiments. Using the...