An Enhanced and Efficient approach towards text detection and extracting text in multi-oriented images

Machine learning is a part of artificial intelligence. It allows a machine to learn on its own and earn knowledge and experience based on the input data set and training set. Text detection and extraction is one of the prominent applications of machine learning. It identifies text in an image with varying orientation, style, alignment, brightness, or contrast in images with complex backgrounds and extracts valuable information from them. Hence solving the problem of reading text in images with text having various variations. The goal here is to help people from different places read the text in languages that the person might not even know. This paper focuses on various challenges in text detection and the techniques used in extracting text in multi-oriented images such as image processing, data mining, or neural network techniques, and a comparative study on them. One such technique is OCR (optical character recognition). This technique helps us to detect and extract text data and convert it into data that can hence be edited and searched. This paper additionally focuses on performance comparison of various techniques that are currently in use. The results of the experimental outcome are judged on parameters like recall, precision, and accuracy.