Investigation of Superpixel Image Segmentation
Aruna Kranthi Godishala, Hayati Yassin, Daphne Teck Ching Lai, Ramaswamy Veena · 2023
Much effort is required to improve the quality of the captured images and to reduce the noise without losing any image features. Thus, image processing is a way to do it, and segmentation is one of the techniques. So far, many researchers have offered numerous segmentation methods which locate the spots of objects and help in the detection of noise. We can also apply a denoising approach to remove the noise. Every approach of segmentation and denoising methods has its own pros and cons. In this work, we perform investigated segmentation and denoising techniques to preprocess the images. We summarize key research findings in the field of image segmentation and denoising to distinguish between the original and processed image, focusing on superpixels or oversegmentation images. The selection of the technique depends on the user, and their target varies from application to application. We have worked on Felzenszwalbs, SLIC, Quick shift, and Compact watershed segmentation techniques and applied non-local means (NLM) denoising techniques to study their performance. A direction for future work is to apply the medical images in identifying the small parts which may help doctors in detecting the presence of any abnormalities.