Abstract :
IR images are widely used many image process applications. Different with visible image, IR images are typically low-contrast, contains blurred edges and a lot of noise, which would bring many difficulties to further analyses. To solve those difficulties, we proposed a new framework for IR image enhancement. The framework not only improves both the global and local contrast of the IR image but also suppress the noise in the IR image. Firstly, IR images are decomposed into a Low-Frequency (LF) component and a sequence of High-Frequency of the input with a multi-scale edge-preserving filter. Then, different strategies are proposed to deal with the LF and HF components. Specifically, to improve the global contrast plateau histogram is applied to the LF component, while to enhance the local contrast a method by using local standard deviation is applied to HF component. To suppress the noise, nonlocal means is employed in detailed images. Finally, the resultant image is reconstructed with the processed LF and HF components. Experiments show that the proposed approach can achieve better performance than other methods in terms of subjective visual effect and objective assessment.
