Handwritten Numeral Recognition Using Support Vector Machine with Bias

Handwriting recognition has always been one of the most challenging tasks in the field of image processing and pattern recognition. As a new classification method based on learning, Support Vector Machine (SVM) is now actively studied and applied widely in pattern recognition problems. In this chapter, we propose an algorithm called SVM with bias, then the proposed algorithm is applied to recognizing handwritten numerals. The experiments on the United States Postal Service (USPS) numeral database demonstrate the effectiveness of the approach.