Automated skin lesion division utilizing Gabor filters based on shark smell optimizing method

In this work, we have proposed an unmonitored method in order to divide the photograph of lesions on the skin using the fabric characteristics. The fabric characteristic in the photograph is described using frequency of energy which is utilized by statistic based approaches called Gabor filter. Optimization of Gabor filters is done by a meta-heuristic algorithm which is named shark smell optimization. Every Gabor filter in the bank is modified to identify the trend of a given frequency and direction in case it is convoluted with photograph of the lesion. The convolving is conducted in the Fourier space. Also the yielded solution photograph is a characteristic which has joined the characteristic vector. Ultimately, the K-means division is utilized in order to distinguish the lesion from the regular part of skin in the photograph. The empirical outcomes indicate that the suggested analytic technique is completely productive in detecting the lesion on the skin for medical purposes. Obtained results demonstrate the validity of proposed optimization approach.