Hearing Impaired or mute peoples are uses sign language to express their thoughts in front of each other as well as normal people. This research paper describes proposed methodology for Indian Sign language recognition where images of alphabets signs are used for recognition. Various image processing techniques have been applied to smooth and filter the images. Similarity index values of testing data and training data have been found as a feature using correlation-coefficient algorithm. This paper consists of a comparison of classification algorithms: Support Vector Machine (SVM), Convolutional Neural Network (CNN), Hidden Markov Model (HMM), and Neuro-Fuzzy(NF) approach. Comparison performs by performance evaluation on MATLAB. Total 200 images of alphabets A to J are tested where 100 images are positive and remaining 100 images are negative. Testing results of positive images consists of accuracy, 94% for SVM, 70% for HMM, 95% for CNN and 97% for NF. Comparison of NF with SVM, HMM, CNN is also describing for the different parameters. Performance is calculated by the confusion matrix where NF approach consists of 96% accuracy.