Towards a standard Bangla PhotoOCR: Text detection and localization

摘要: A complete Bangla PhotoOCR requires a series of carefully chosen algorithms. Text extraction from images is long-standing active research area. It even more attractive today due to the availability low-cost mobile image acquisition devices. Many researchers have addressed this problem using different approaches. Often times, first step towards text detection areas. texts, specially in images, pose unique set challenges than texts other languages. In paper, we experiment with two established approaches, available for languages, automatically localize complex natural scene developing system. our approach, features are extracted an wavelets based decomposition and histogram calculation techniques. We use 56 train types classifiers (ANN SVM based) images. Our experimental results show that ANN good classifier identifying texts.