Improving Optical Character Recognition(OCR) Accuracy using Multi-Layer Perceptron(MLP)

Abstract

Optical Character Recognition (OCR) is a cutting-edge application that has been made possible due to advances in technology. Optical Character Recognition involves using deep learning algorithms such as Multi-Layer Perceptron and Support Vector Machine to create a system that can recognize characters in an image. The Optical Character Recognition system first segments each character individually as part of the pre-processing step, after which it post-processes the image to compare each character to the pre-processed data. In the modern world, Optical Character Recognition is a crucial technology used in many industries to shorten turnaround times while increasing accuracy. However, one of the challenges faced with Optical Character Recognition is that it takes longer to match exact results and produces results with lower precision and character misreadings. To address this challenge, the pre-processed data used by Optical Character Recognition systems now includes practically all fonts. This means that the Optical Character Recognition system has to match the characters with a much larger dataset, making the process more time-consuming and less accurate. To improve the accuracy and speed of Optical Character Recognition systems, several methods have been employed. For instance, post-processing techniques such as error correction algorithms can be used to detect and correct errors in the Optical Character Recognition output. Additionally, the use of deep learning techniques such as Conventional Neural Networks has been shown to improve the accuracy of Optical Character Recognition systems. By employing these methods, Optical Character Recognition can become an even more powerful tool that can shorten turnaround times while increasing accuracy.