Characters And User Interface Components Recognition From Desktop Digital Images

This paper explores the content extraction from digital images, specifically from the desktop application digital video images. Content like text and user interface (UI) components are extracted and recognized. An algorithm has been developed for text and UI components segmentation and recognition. An artificial neural network (NN) based optical character recognition (OCR) technique has been developed, which transform the character image into plain text. Ul components segment is recognized based on simple pattern matching technique Text and Ul components contained in such image can be seen as an intermediate stage between image processing procedures and domain knowledge. The algorithm has been tested on images recorded from desktop application screen shots; the experimental results highlight the quality of the solution proposed.