Design And Implementation Of System For Air-borne Electronic Image Stabilization

Electronic image stabilization is a new technology to stabilize video image sequence. Because of its advantages of high image stabilizing precision, flexibility of application, it becomes one of the hot subjects in video processing field. Caused by the unsteady or moving platform mounted on plane, the change between the frames of image sequence is large, which influences the use of the information of image. The EIS technique can eliminate or alleviate the influence on the image sequence caused by the moving carrier and improve the quality of the image greatly. We have carried out an intensive study on the EIS technique of air-borne optics imaging system, and put forward an effective method for realizing stabilization of the air-borne video image and realized an algorithm based on cross-block gray projection algorithm and Kalman filter motion compensation. We also accomplished the hardware design and realization of a high-performance hardware platform of EIS. The main research contents of the thesis are as follows:We expatiate on the principle and method of electronic image stabilization in detail, configuration of the system and key technique of realization. We also build the video image model used for image stabilization and puts forward the method and standard of the evaluation for EIS based on the study of the motion estimation and compensation.According to oscillation of airborne imaging system and high real-time requirement of airborne EIS, an algorithm based on cross-block gray projection algorithm and Kalman filter motion compensation is proposed. This algorithm uses histogram proportion technology to enhance the image and chooses four blocks to estimate the local motion. The general motion is calculated based on the model of image motion. The experimental results show the validity of presented method for compensating the rotation and translation motion, which can realize "real-time" processing to airborne image sequences. The RMSE is smaller than 1.We use the motion compensation technology based on Kalman filter. It uses Kalman filter to wipe off the random motion and keep the essential motion of the image sequence. Then we get the value of compensation by finding the difference between the result of the Kalman filter and the motion estimation of the image sequence.According to the requirement of'real-time'and the algorithm flow for airborne EIS, we adopt a hybrid hardware configuration based on FPGA and high speed dual-DSP processor as the design scheme and accomplishes a high-performance image information processor. The size of this high-performance image information processor is 14cm*14cm*1cm.It has great ability of data processing and memory, high dependability and its power consume is less than 7W. It can satisfy every requirement of the application environment on plane.Based on the platform, the image sequence taken by the air-borne imaging system is tested in the EIS experiment. The experimental results show the validity of the algorithm based on gray projection algorithm distributed crisscross and the high performance of the image information processor. The system can realize stabilization of the air-borne video image by 25 frames per second.