There is a severe contradiction between the 14bits output data of the infrared thermal imager and the 8bits input data of the monitor in a real-time infrared image processing system. A novel algorithm for real-time infrared image enhancement based on FPGA is proposed not only to solve this problem but also satisfy the real-time and mass data need of the system. First

an image mean filter is used to compute the local limit values for each frame dynamically. Next

a multi-level average statistics method is designed to calculate the values of segment points for each gray normalization partition. Finally

the means of different linear normalization on different gray level segments is taken to enhance the infrared image. The whole algorithm is developed on the FPGA hardware with Verilog language. Experimental result shows that this algorithm settles the problem that the output values of un-cooled focal plane array detector are varied with its working duration. Moreover

the algorithm could reduce the impact from little bad pixels with very large values

and enhance the details of an image which has some objects of high and low temperature simultaneously in a scene. Compared with histogram projection

one of the most traditional image enhancement methods

this algorithm shortens the computation time about 1.38ms and consumes only 1/42 memory resource than histogram does. Therefore

it could be implemented by FPGA easier and perform better in a real-time infrared image enhancement system. In terms of engineering application

the proposed scheme with simple architecture

high reliability and low cost characters could provide a valuable reference for other design of a real-time infrared image processing system.