Abstract

There are various approaches of Hough transform computation proposed in the literature. All of them work with the analog expression of straight line and map the evidence collected in in parametric domain into bins. The bins with high accumulation correspond to the straight lines in the image. This talk will describe the limitations of this general approach and propose a new way of computing the Hough transform. The main idea here is to consider the Digital representation of Straight Line (called DSL) and generate masks for DSLs for lines with different orientation and origin. The object pixels accumulated on the masks are put in the corresponding bins. The bins with highest scores are considered as candidates for the line. Then the existence of line is confirmed by continuity and other Psycho -visual considerations. Advantages of the approach and some application to document images will be described.

 

Biography：

Professor Bidyut Baran Chaudhuri is the Head of Computer Vision and Pattern Recognition Unit of Indian Statistical Institute, Calcutta and a distinguished J. C. Bose Fellow(2010-). Professor Chaudhuri received his B.Sc (Hons), B. Tech and M. Tech degrees from Calcutta University, India in 1969, 1972 and 1974, respectively and Ph.D. Degree from Indian Institute of Technology, Kanpur in 1980.He is a fellow of IEEE, International Association for Pattern Recognition (IAPR), Indian National Sciences Academy (INSA), National Academy of Sciences (NASc), Indian National Academy of Engineering(INAE), Institute of Electronics and Telecommunication Engineering (IETE), West Bengal Academy of Science & Technology, Optical Society of India, Society of Machine Aids for Translation etc.

Prof. Chaudhuri is an Associate Editor of International Journal of Document Analysis and Recognition, Int. Journal of Pattern Analysis and Machine Intelligence, Int. Journal of Computer Vision as well as VIVEK. He also served as Guest Editor of several journals.