Wide Dynamic Range Vision for Robot Eyes based on Double Capturing Scheme

A new method of combining double captured images to extend dynamic range is proposed in this paper. Two images taken with long and short exposure times respectively are combined by proposed DSPA (difference & slant based pixel adjustment) algorithm, dynamically varying weighted averaging scheme to present a scene of wide dynamic range. Compared with the previous works using multiple capturing, our approach captures only twice to facilitate practical implementation. It is characterized by simple arithmetic operations which can be easily implemented by a small H/W. Compared with the previous double capturing scheme, it shows enhanced contrast which contains more edge information, easy control of resultant effects while still retaining simple architectural merit. The test results for the several scenes show it has logarithmic convergence characteristics and more edge quantity compared to the previous works.