Image edge detection using objective function and fuzzy C means

Images contain information based on their color intensity. By observing the degree of color intensity difference between two pixels or more, it can be determined an edge of image. The problem encountered is that if the color intensity difference between the pixels that is assumed as an edge is not significant, so the edge determination of image becomes unclear. An objective function can be used for calculating the magnitude of 4 direction values (horizontal, vertical, and 2 diagonals) of a pixel in the image. The result of this calculation can be used as feature of image texture. By analyzing the characteristics of image texture features, they can be grouped to determine whether the pixels are included in the category of background, edge, or noise. In this research, the image texture features clustering are done by implementing Fuzzy C Means algorithm based on the data distribution of mean and standard deviation values of each 4 magnitude direction values of a pixel which have been calculated based on the objective function. The value of the cluster centers obtained from the data clustering is further ranked to know their differences. Based on analysis, this method can distinguish 3 image texture features clearly (background, edge, noise). Therefore it can be concluded that cluster center grouping with the largest mean value can be used to form an edge of the image.