Impulse Noise Removal from Digital Images by Using Edge Preserved Morphological Mean Filter
Shehin Shams P., Vipin Kumar K. S · 2018
An edge preserved morphological mean filter oriented image restoration system is proposed. It targets the images excessively corrupted by impulse noise. Averaging filter is the usual choice for this purpose. However, it is desperate for disposing of excessive-density noise. Most of the methods do not give importance to edge preservation. The proposed restoration technique comprises two phases. Those are identification of corrupted pixels and reconstruction of those corrupted pixels. Usually a center sliding window is used for this purpose. An absolute directional difference of the pixels is used in the first phase of the method to recognize the pixel values influenced by shot noise. For this purpose an edge preserved model using a Gaussian kernel is followed. An edge prediction technique is also used. In next stage, the morphological dilation operation is used to replace the neighborhood noise pixels. Also, this technique is applied to both color and grayscale images and analyses their efficiency.