Detail Enhancement and Dynamic Range Compression for High Dynamic Range Infrared Images

This paper presents an image enhancement method for the effective visualization of High Dynamic Range (HDR) Infrared (IR) images on Low Dynamic Range (LDR) display. Recent IR sensors are capable of producing raw images with pixel intensity values represented by 15 bits, whereas the sensitivity of most of the display devices is limited to 8 or 10 bits, which stems from the fact that human eye can sense only about 128 gray levels in an image. Hence, rendering real world IR images on state of art display devices requires compression of HDR IR images without losing the perceptibility of details. This technique is termed as Dynamic Range Compression (DRC). Standard DRC methods have achieved moderate success in faithfully reproducing a HDR image on a LDR display, but they lack flexibility in manipulating the minute details, which may contain very important information in certain scenarios. This paper proposes a new method, wherein detail enhancement and dynamic range compression are treated as two separate entities. It employs a preprocessing module involving bilateral filter to enhance the details in the input image. The detail enhanced image is then mapped to the display dynamic range using Histogram Projection. Experimental results show that the proposed method significantly enhances the image details and improves the overall contrast.