A vision-based hand gesture identification system is developed to recognize the hand shapes from the Indian Sign Language (ISL) - 26 letters A-Z. There are several similar looking gestures in this signage system. Thus, considering these in common, for the system recognizing alphabets, a total of 26 signs are used for identification. The main feature masked hand feature was extracted from the sign images to act as the feature vector using HSV modelling. The system performance for sign recognition is measured by considering these features independently as the feature vector for recognizing the ISL alphabets. The effect of combination of the features such as HSV colour selection and the classification algorithm such as CNN, the system was able to detect the various signage easily. Also, the system performance is measured by extracting the same features by creating skin masking image of the signs of alphabets for ISL alphabets recognition.