Infrared Image Enhancement Based on Local Entropy-Local Contrast and Dual-area Histogram Equalization

We propose an infrared image enhancement algorithm based on an improved local contrast (LC) significance detection algorithm and two-area histogram equalization to improve the visual effect of an infrared image, highlight detailed information, and suppress noise. First the LC saliency detection algorithm was combined with local entropy weighting to obtain the saliency map. Then, the saliency map was adaptively segmented into foreground and background regions by the K-means algorithm. Finally, the foreground sub-histogram was equalized using a local variance-weighted distribution. The background region was enhanced using contrast-limited adaptive histogram equalization. Experimental results showed that the subjective effect of the algorithm in this study was better than the current mainstream algorithms, and the objective evaluation parameters, such as peak signal-to-noise ratio, structural similarity, and entropy, were also improved.