A fuel rod is a fundamental unit of a fuel assembly, and it directly impacts the safe operation of nuclear reactors. To efficiently detect internal defects in fuel rods, a high-resolution visual nondestructive testing method, X-ray imaging, i.e., digital radiography (DR), is employed.

This study aims to address the issue of low contrast in fuel rod X-ray DR images by proposing a brightness fusion and multiscale optimized enhancement algorithm.

First, logarithmic and gamma transformations and further refined by incorporating local information fusion were employed to correct the brightness of fuel rod DR image. Subsequently, a wavelet function was applied for multiscale decomposition, enhancing and sharpening low-frequency components with Retinex, and non-local means (NL Means) was applied to filtering high-frequency components. Then, image enhancement was realized via wavelet reconstruction. Finally, quantitative analysis experiments were conducted using the DR images of fuel rods to evaluate the performance of the algorithm by means of two representative image quality assessment metrics, i.e., average gradient (AG) and information entropy (IE), and compared with that of new low-light image enhancement (NLIE) algorithm, homomorphic filtering (HMF) algorithm, and low light image enhancement (LIME) algorithm.

The experimental results demonstrate that quality of fuel rod DR image is significantly improved by image brightness fusion and multiscale optimized enhancement algorithm proposed in this study with the highest information entropy (IE) of 6.834 5, which is 10.2%, 3.3%, and 12.6% higher than NLIE, HMF and LIME algorithms, respectively, hence the internal defects of fuel rods are better highlighted.

This algorithm proposed in this study not only effectively improves the overall and local contrast of the fuel rod DR image, but also significantly highlights edge details, verifying its effectiveness in improving the quality of X-ray DR images.