Handwriting Digit Recognition using United Moment Invariant feature extraction and Self Organizing Maps

Handwriting Digit Recognition (HDR) have a high level of research difficulty, because handwriting forms are not consistent and always changing due to a distortion. So, the accuracy HDR is significant in many areas such as recognizing the postal codes in the cover letter and customer account number during banking activities. To solve this problem, this research will develop a recognition that use United Moment Invariant feature extraction and Self Organizing Maps for recognizing the actual digit. The dataset that will used is MNIST which contains 10,000 images of digits from 0 to 9. Since many researches proved that Self Organizing Maps can produce is very good performance.