Using Morphological Operators and Inpainting for Hair Removal in Dermoscopic Images

    June 2017
    Julie Ann A. Salido, Conrado Ruiz
    Image of study
    TLDR The new algorithm removes hair from skin images better than previous methods, helping diagnose melanoma.
    The document describes a new algorithm for removing hair from dermoscopic images, which is important for melanoma diagnosis. This algorithm uses image processing techniques such as median filtering, morphological operations, and harmonic inpainting. It was tested on the PH² dataset and showed better performance than the existing DullRazor software, with a higher Peak Signal-to-Noise Ratio (PSNR) of 33.41 compared to DullRazor's 32.44. Although the algorithm had some failure cases, it generally resulted in fewer false positives and is considered effective for hair removal, which is a crucial pre-processing step in skin lesion analysis. Future improvements could include more advanced hair detection and image completion methods. The document, however, lacks specific numerical results or the number of images tested to fully assess the study's strength.
    Discuss this study in the Community →

    Related Community Posts Join

    6 / 1000+ results

    Similar Research

    5 / 1000+ results