Infrared Image Sequences. Plaque Target Detection And Realization

In recent years, with the rapid development technology of small target detection in infrared image, people already have full knowledge about the important of the technology. But with the characteristics of small area, far distance, weakness of shape, lose of detail of image and low SNR etc, the infrared small target detection technology is one of the key and difficult point of infrared guidance system. The research of infrared small target detection is still an issue with great challenge.This paper first analyzed the common algorithms of image pre-processing and target detection, concluded the advantage and shortcoming of different algorithms,then, in connection with the characteristic of the target and background in this project, brought forward a infrared small target detection method. First, we expound an infrared small target detection method in single frame infrared image, which connection with mathematical morphology and detect the connected domain in small region. But, because the own characteristic of the small target in this project, this method can not detect the small target in every single frame infrared image. Integrating the characteristics of correlation between the two adjacent frames image in sequence image, we expound a method of small target detection in sequence image, which method integrates the technology of infrared image registration. Based on this method we study the SIFT algorithm in infrared image registration. And the Matlab simulation shows that this method can raise the success rate of detecting the small infrared target and reduce the false alarm rate. This method meets the requirements of the project in a certain extent.In this paper the real-time image processing system realizes complex algorithms using the structure of FPGA+DSP. In order to get the most efficiency, the image processing algorithms are classified. As its core of image processing system, FPGA and DSP play their respective strengths. For the high real-time algorithms which realize easily and need quantity operations can be realized use of high performance FPGA, it can save the hardware resources and increase the processing speed. The control of image recognition and registration algorithms are complex, require high computing capacity and need much more multiplication and addition operations, posing a high demand over system performance. DSP ideally meets the requirements of the system due to its unique structure and excellent performance. This divide enables the real-time image processing system.Study indicates that image recognition system based on DSP+FPGA structure is more flexible and applicable to general purposes, and implements modular design, thus increasing the overall efficiency of the algorithm. By fully leveraging the combined advantages of DSP and FPGA, the system could enable accurate and fast image recognition. With flexible configuration of software and hardware, the system could have most image processing functions and applicable in the fields of industrial visual inspection and automobile license plate recognition etc.