Pipeline Implementation of New Segmentation Based on Cognate Neighborhood Approach
A. Nagaraja Rao, U. S. N. Raju · 2008
In this paper, we cast the segmentation problem as the maximization of cognate information between the row and column pairs of the neighborhood. For this, the maximum and minimum gradient on pairs of row and column of the local neighborhood are calculated and applied on the original image. The maximum and minimum row and column pair of the local neighborhood forms the contribution of primitive morphological operations, dilation and erosion respectively. For extracting strong edges edge increase and edge decrease are applied on local neighborhood. The final segmentation is obtained after applying the above preprocessing steps by using a new approach of cognate neighborhood. To test the above process of segmentation the method is applied on brodatz textures, leena and brain MRI images, which resulted a good segmentation.