Image Denoising: An Overview of Important Methods in Spatial and Transform Domain

Images play an important role in our day-to-day life. Image finds its use from simple documentation up to complicated medical and surveillance applications. There is an increased demand for good-quality images. Digital cameras use light sensors for capturing images. The gray-level value of image is the result of number of photons reaching the sensors. The electrical voltage equivalent to the light intensity is then converted into digital. However, due to atmospheric conditions, improper lighting, camera sensors, type of receivers, transmission channel, etc., images get corrupted by noise. Reduction in the quality of image affects the performance in several applications such as segmentation, traffic surveillance, object recognition, tracking, etc. Denoising is therefore a prerequisite in most of the applications. Denoising can be achieved in transform domain, spatial domain, and also using hybrid methods. The main idea behind denoising is to smooth the image that can suppress the noise and on the other hand, smoothing also decreases the image details. Hence a good equilibrium between smoothing and preserving image details is very much necessary. This chapter highlights important methods both in spatial and transform domains that can achieve denoising. Some of the hybrid methods are also introduced. The assumption here is of image with Gaussian noise.