The objective of noise removal is to detect and remove unwanted noise from a digital image. The difficulty is in determining which features in an image are genuine and which are caused by noise. In general, it is assumed that variations in intensity and colour will be gradual in an image, so points which are significantly different from their neighbours can often be attributed to noise. Hence, the central idea behind many noise removal algorithms is to replace anomalous pixels with values derived from nearby pixels. Local averaging is commonly used for this purpose, which has the side effect of smoothing the output image. Many noise removal algorithms have parameters which can be adjusted to trade off noise level versus smoothing, so the ideal image for subsequent processing can be interactively selected.