Fpga-based Real-time Image Processing System Design And Algorithm Research

Infrared technology is widely used in both military and civilian fields to scan, detect and identify target. The development of infrared focal plane array technique is crucial for the performance of infrared device, and it is important to develop the low price, small bulk, low power consumption and high performance infrared thermal imaging device. However, currently the imaging quality is always not high enough because of the limitation of infrared detector. Non-uniformity correction is indispensable in IRFPA device before the practical application. After the non-uniformity correction processing of infrared image, there are some defects in the images. The dynamic range of the gray image is too small and the image details are not clear enough. The image enhancing algorithms are widely used in infrared imaging processing field to improve the image quality.In this dissertation, the image enhancement algorithms and video display are concentrated on, modified and improved. A platform for implementing the algorithms is designed and realized. The main contents are as follows:(1)According to the theory of infrared imaging technique, a mini type infrared thermal imaging system, which contained the voltage bias, sequence drive, A/D, temperature controlling, NUC, video transformation , has been designed in this article based on FPGA. two SRAMs are used in the system for ping-pong operation to adjust the data transmission rate between the front data collection module and display module. The system can operate at a rate of 70 frame/s for 320×240 arrays.(2)In the image processing module, non-uniformity characteristics of infrared focal plane and the characteristic of infrared image are studied. Hardware implement circuits for two-point non-uniformity correction algorithm, linear transformation algorithm for gray image and pseudo-color algorithm are proposed. Histogram equalization algorithm, linear smoothing filter algorithm and nonlinear smoothing filter algorithm are simulated using MATLAB. The result shows that histogram equalization algorithm can enlarge the dynamic range of gray image, and the smoothing filter algorithm can reduce high-frequency noise. The improvement based on the traditional pseudo-color algorithm is proposed and a new adaptive nonlinear pseudo-color algorithm is used to make the image look soft and different temperature regions and the target can be easily recognized.