A Novel Technique for Handwritten Text Recognition Using Easy OCR

Among computer vision's many new subfields, character recognition is one of the most promising. Humans' superior recognition skills make them the most powerful species. Humans have little trouble deciphering the hand transcription. There are distinct patterns to look for in various languages. The text is easily recognizable by humans. The machine cannot recognize the manual transcription. The system needs help recognizing the text. Processing the input image, extracting features, and training a classification schema are all part of the text recognition process. The computer is taught to look for similarities and differences between different handwriting samples in this method. The program reads a picture of a handwritten transcript and creates an electronic version. Today, cutting-edge scientific methods expand humanity's horizons in all technological endeavors. Character recognition, often known as OCR (Optical Character Recognition), is one example of such a field. The approach presented here is built on EasyOCR and Regular Expressions, which facilitate exporting trip sheet information to CSV. The suggested model divides the image into many boxes based on the intersection of lines and then looks for individual characters within each box. Recognized texts are saved independently, allowing for format conversion at a later date. Unlike other types of OCR (Optical Character Recognition), which focus on recognizing characters generated by a computer or typewriter, this project focuses on developing an algorithm for recognizing handwritten characters, commonly known as HCR (Handwritten Character Recognition). An innovative method, based on Convolutional Neural Network (CNN) and including character feature extraction approaches, is proposed for recognizing English language characters. Over 90% of the time, CNN correctly identified the characters being shown.