Visual Attention In The Infrared Image Processing Application

The main trend in infrared guidance's development is infrared imaging guidance. And it's a main branch of precision-guided technology's development.The main constrained factor of infrared imaging guidance technology development is the contradiction between large field of view, multi-pixel and real-time operational capability.The high efficiency of human eye's information obtain comes from its attention mechanism. So the study on vision system bionic technology and its application on infrared imaging guidance is important. And the technology may be very useful to the design of the guide head structure, the infrared target image information obtaining and processing and the target recognition. The mathematic model of biological vision system's information obtaining and processing mechanism and computer emulation on the model are the main study tasks. The research productions not only can be used in precision guidance of infrared imaging guidance missile, but also be used in many kinds of infrared prospecting apparatus and robot's vision system. The study has favorable practical meaning and well prospect.Firstly, the study achievements of the vision system channel, vision nervous system and information processing are introduced based on the vision physiology and psychology. The eye's vibration and attention shift are also given. Based on these achievements, three levels of the attention are proposed and it is the physiologic base of vision attention mechanism.The transfer operation is classified according to the diversity of demand and task. Based on the log-polar transformation(LPT), two algorithms, the forward one and the reverse one, are summarized. The data size of the forward algorithm's result matrix is small. But the transfer matrix's front-end is discrete and it's disadvantageous to the subsequent process.The filling operation is considered in the reverse algorithm. So the transfer matrix's full image can be obtained.According to the above-mentioned forward algorithm and the reverse one, the scale change in the biological vision system's visual cortical layer is explained. The mathematic model graph corresponding to the vision system is also given.The scale invariability and the revolution one of the classical LPT are discussedin the paper. An example is done to explain the two features.But the classical LPT hasn't the translation invariability. So the Fourier transformation is used to solve the shortcoming.The variable parameter log-polar transformation(VPLPT) is proposed. Using the new algorithm, the central highly resolution area's size can be adjusted. The data size is also reduced. So it's advantageous to real-time operation.The data size comparing analysis is done between the classical algorithm and the new one. The results shows the VPLPT's rapidity.The new algorithm is used in a helicopter's infrared image processing. The simulation also shows its validity.The framework of early vision model with selective attention mechanism is proposed. The model is also refined.Aimed at the characters of the infrared imaging guidance, a application scheme in the infrared image process based on the vision model is given.The FPGA implementation scheme of LPT is also discussed in the paper.