MATLAB-Based App for Handwritten Numeral Recognition

Handwritten numeral recognition (HNR) is widely used in various industries and is essential to the research and development of optical character recognition (OCR) technology. Given that most of the current HNR algorithms require high-quality input images rather than naturally taken photographs, a MATLAB-based application with a graphical user interface (GUI) is designed for pre-processing and segmenting photographs of handwritten numerals and then recognizing the numerals. For the various conditions that a naturally taken a photograph may have in practice, the application can perform denoising, perspective correction, and binarization of the original image, as well as removing horizontal lines, grids, or boxes from the background. The application can then segment and recognize every numeral from the processed image. It has been experimentally verified that the application can satisfactorily pre-process the original image while providing high accuracy in recognizing numerals.