A novel modified histogram equalization approach for enhancement of non-uniform illumination of Chest X-ray images

This research paper presents a novel technique for enhancing non-uniformly illuminated chest X-ray (CXR) images. Existing methods based on histogram equalization (HE) often produce unnatural appearances and washed-out effects due to over-enhancement, leading to uneven illumination across the image. To address this problem, the proposed technique divides CXR images into two exposure areas: underexposed (UE) and overexposed (OE), using an exposure parameter. Each sub-area’s histogram is transformed using a nonlinear weight derived from the histogram’s cumulative density function (CDF). Next, the histograms are modified to be equalized using different ‘intensity expansion’ and ‘mapping directions’ for ‘UE’ and ‘OE’ sub-areas. The proposed method has been evaluated using the Kaggle COVID-19 Radiography datasets, which consist of images categorized into four groups: COVID-19 (3616 images), Lung Opacity (6012 images), Normal (10192 images), and Viral Pneumonia (1345 images). The technique was compared against three state-of-the-art methods: Histogram Equalization (HE), Image Subdivision and Quadruple-clipped Adaptive Histogram Equalization (ISQCAHE), and Gamma Correction for Brightness Preservation (GCBP). Parameters applied for quantitative analysis are Discrete Entropy (DE), Mean Square Error (MSE), Peak Signal Noise Ratio (PSNR), Absolute Mean Brightness Error (AMBE), and Image Contrast Factor (ICF) and so on. The proposed technique demonstrates impressive improvements in quality image, illumination, detail preservation, and naturalness in its results. This technique specifically achieved greater values for DE and PSNR and lower values for MSE and AMBE. It enhanced ICF compared with existing methods, indicating high performance in enhancing non-uniformly illuminated CXR images.