Performance Analysis of Image Denoising Techniques in Skin Cancer Detection

Because visual information is increasingly being sent in digital image format, identifying noisy data becomes a prevalent difficulty in many research and application sectors. To reduce noise while maintaining image information, many noise reduction techniques have been developed in the modern period. Researchers always struggle with the challenge of noise removal from original image. This study proposed a pre-processing method to improve image quality in terms of performance parameters, making the subsequent image processing work in the diagnosis of any type of skin cancer easier. The major focus of this work is on image denoising or filtering, which improves image quality for subsequent image processing processes in the diagnosis of normal vs. malignant skin. Objectives: To calculate various denoising performance parameters like peak signal to noise ratio (SNR), mean square error (MSE), normalized root mean square error (NRMSE), and others. Methods: In this research work, image denoising technique was performed by using different types of noises and filters. Statistical Analysis: In this research work, we found observational outcomes after noise addition to skin images in terms of pixel value, darkness, brightness related to each pixel. Gaussian and median filter gives MSE between 0 and 17%, NRMSE ranges between 0.025 and 0.425, and PSNR is obtained from Min 15 dB to Max 40 dB.