Research On Infrared Image Detail Enhancement Algorithm And Its FPGA Implementation

Comparing with a 8 bit width picture taken by visible rays, a infrared picture taken by thermal infrared imager has 14 bit width and contains more information, it’s more accurate and sensitive. But common screen is 8 bit width, infrared pictures must display on screen via dynamic compression. As a result, infrared pictures will lose a mass of details if the dynamic compression is not appropriate. When compressing a infrared picture, the way to protect its detail is particularly important. This article aims to study on how to enhance detail when dynamic compression..The feature of infrared pictures and the development tendency of detail enhancement of infrared pictures are introduced. Some common methods of detail enhancement of infrared pictures and each advantage are discussed. For the detail loss via dynamic compression, two ways based on unsharp masking are presented. One bases on Bilateral Filter, the other bases on Guided Image Filter. Algorithm basing on Bilateral Filter uses a self-adaption Gaussian Filter to remove Gradient inversion, uses Plateau Histogram Equalization (PE) to compress base frequency, enhances the detail to depress noise with a self-adaption coefficient, and combines the base layer and the detail to gain the enhanced 8 bit output. Algorithm basing on Guided Image Filter improves the former, uses Guided Image Filter to smooth pictures for it can avoid Gradient inversion, presents a new coefficient to enhance the detail and depress the noise. When the noise of infrared pictures is to large, a method is presented to depress noise with edge protection and detail enhancement. The matlab simulation provides a good quality of noise reduction and detail enhancement comparing with normal method.The ways of realizing PE, Bilateral Filter, Guided Image Filter on hardware are presented. By using XILINX FPGA of Spartan6, a real time system is established which contains infrared videos collection, detail enhancement when dynamic compression, result display on a screen. The system shows a good result on detail enhancement.