Image Query-Based Tablet Identification System by Examining Various Text Recognition Classifiers

Classification is the most important and frequently used technique in image processing. Detecting text region in images is helpful in computer vision applications, like searching, analyzing, and retrieving image. Text detection and extraction (TDE) is an important step of image query-based tablet identification system, as it is used as an important feature during tablet identification. The key purpose of this work is to improve the effectiveness of text detection and extraction process. This research work deals with the approach based on a binarization method that uses canny edge detection, Otsu thresholding, enhanced connected component labeling with automatic threshold procedure. Experimental results show that the support vector machine (SVM) classifier is efficient than the other three classifiers with respect to accuracy, speed, precision, recall, and F-measure in both consumer and reference images.