Document Image Text Extraction Using Emperor Penguin Optimization with Convolutional Neural Network

Text attained from natural scenes different information: accordingly, it is greatly utilized in different applications to understand the scenarios of image and also to retrieve the visual information. Human beings greatly benefit from the scene image's semantic content, which helps to understand their surroundings. The process of text identification and character recognition is more difficult in these natural images since the text presents a flexible appearance in an unrestricted environment. Therefore, implemented an Emperor Penguin Optimization (EPO) with Convolutional Neural Network (CNN)-based deep learning process is employed in this framework to effectually detect the text and to recognize the character from the scene images. To eliminate noise natural scene images, a grayscale conversion and image noise removal is typically applied during the preprocessing step. The DWT, LBP, and ResNet 50 techniques are used in feature techniques process, which are assists for the classification process. Following that, extracted features are selected by EPO technique. At last, with these selected features, the character recognition and text detection are achieved successfully by CNN techniques.