No abstract

An ordinary FCM algorithm does not completely utilize the spatial information in the image. In this paper, we exhibit a fuzzy c-means (FCM) method that integrates spatial information into the membership function for clustering. The spatial function is a summation of membership function in neighborhood of every pixel under consideration. In image segmentation fuzzy C-means (FCM) clustering method making it effortlessly traps into local optimum and huge calculation, image segmentation algorithm based on the modified centroid weight particle swarm optimization (MPSO) and Spatial FCM clustering algorithm is proposed. This technique is a powerful method for medical image and multimedia image segmentation with spatial information. The simulation outcomes and the comparison among the proposed method, FCM and K-means algorithm indicate that the proposed method can achieve better segmentation and excel the existing FCM algorithm in several performances, such as the average error and cluster centroid.

No abstract

In this paper, an efficient feature extraction method based on the Kande-Lucas-Tomasi (KLT) using fast independent component analysis (Fast ICA) & Anthropometric Model as the distance measure is proposed. Each face is extracted facial organs are marked for Anthropometric Model (AM) distance measure. The KLT facial coefficients of low & high frequency in different scales & various angles are obtained. The coefficients are utilized as a feature vector for further processing. Considering the extracted face image & adopt the Fast-ICA algorithm based on entropy to extract the face feature information. Finally, according to the Anthropometric distance to classify face feature & Artificial Neural network (ANN) used to estimate age for all kinds of facial databases. Experiments are done by using the YALE & FERET databases. An experimental outcome shows that the recognition rate Mean Absolute Error (MAE) of the proposed algorithm is acceptable & very promising, & confirm the success of the proposed face feature extraction approach.

No abstract