Gradient Based Histogram Equalization of Thermal Infrared Images

For the past 20 years, histogram equalization (HE) has been used in image enhancement as the preferred way to increase the contrast. Thermal infrared (IR) image enhancement techniques primarily utilize HE because of the limited pixel variation in IR images. There are two types of HE, local-mapping and global mapping. New methods based on HE are continually being developed and improved. The recently developed gradient-based histogram equalization (GBHE) is one such method. GBHE, a global-mapping method, utilizes gradient operators in a preprocessing step before applying the HE to the image. The original experiments for GBHE were limited to a handful of filters and grayscale images. This thesis proposes to use GBHE and improve the method for IR image enhancement. More operators, such as Michelson and Weber visibility operators, and more filter sizes were tested and shown to improve the GBHE performance for IR images. The GBHE method performance was compared to two methods classical HE, a global-mapping method, and contrast limited histogram equalization (CLAHE), a local-mapping method. The GBHE method has better performance over HE and slightly worse performance when compared to CLAHE.