Underwater Prefabricated Structure Detection Image Degradation Classification Model Based on HSV-LAB Color

Zhu Taoguo^{（Correspondence）}  Shen Xiaoya  Jiang Bianyu  Guo Jiangpeng  Bao Huilin  Li Xuedong  Guo Lili

School of Civil Engineering, Lanzhou Bowen College of Science and Technology, Lanzhou, Gansu Province,P.R.China，730100；

Abstract：To address the three typical degradation phenomena—color cast, low luminance, and blur—frequently encountered in underwater images of prefabricated structures captured within complex subsea environments, this study proposes a multi-feature underwater image degradation classification model grounded in the joint analysis of HSV and LAB color spaces. The pipeline first preprocesses each image by converting the native RGB representation into both HSV and LAB color spaces while simultaneously performing noise filtering. Subsequently, three discriminative features—namely, the standard deviation of the HSV hue channel, the mean value of the LAB lightness channel, and the Laplacian variance of the grayscale image—are extracted as the core classification descriptors. Empirically determined thresholds are then applied to these features to enable quantitative discrimination among the three degradation categories. The proposed model was validated on a dataset comprising 30 underwater inspection images of prefabricated structures. Experimental results demonstrate that the model can rapidly and accurately distinguish the degradation type of each image: 53.3 % of the images exhibited color cast, 23.3% suffered from low luminance, and 23.4% were assessed as free from noticeable degradation. These classification outcomes furnish a precise pre-enhancement diagnosis for subsequent scene-specific image restoration. Moreover, the average processing time per image is below 0.01 s, satisfying the real-time requirements of practical subsea inspection tasks.

Keywords: underwater prefabricated structure inspection; image degradation classification; HSV color space; LAB color space; Laplacian variance

References

[1]ChenHaiwen.Underwaterimagequalityimprovementmethodbasedonlightscatteringsuppressionandlocalcontrastenhancement[D].HefeiUniversity,2024,DOI:10.27876/d.cnki.ghfxy.2024.000081

[2]Opticalpropertiesofthe“clearest”naturalwaters[J].Limnology&Oceanography,2007,52(1):217-229.DOI:10.4319/lo.2007.52.1.0217