AUTOMATIC DETECT OF HAND WRITTEN ABNORMALITIES FOR SINGLE-DIGIT NUMBERS, USING ARTIFICIAL NEURAL NETWORKS ON ANDROID SYSTEMS

Mathematics disorder is inability of individuals in skills related to calculation according to intellectual capacity and level of education .Mathematics disorder is divided in to four categories of skills: Language skills (such as comprehension and naming mathematical terms and concepts), perceptual skills (such as data classification and recognition, and etc.), mathematical skills (such as counting objects, etc.), and skills related to attention such as (copying and memorizing digits). It can be stated that, mathematics disorder exists among some children who are not mentally retarded. Developing efficient methods for recognizing handwritten identifier can be helpful for automatic recognition of listed letters and numbers in forms, checks values, and a considerable number of other applications .The biggest problem in this regard is diverse ways of drawing identifiers .One of the initial issues raised in neural networks as a solution is identifier recognition. It is a method for recognition of single-digit numbers according to diagnosis based on main body of the number .This is similar to what human beings do while writing numbers. Thus, the coordinates of each of the parts of the body are identified and recognized using the x, y coordinates. Using more data for the neural network training phase led to improved recognition of numbers in comparison with the previous phase. Increasing the number of samples and tests on different images (low quality, with margin and angle), we are able to improve part of the code which relates to the removal of margins, etc. and enhance system performance.

artificial neural network, Preprocessing, PC filter.