Reliable Dual-Band Based Contour Detection: A Double Dynamic Programming Approach

Finding contours in constrained search space is a well known problem. It is encountered in such areas as tracking objects in videos, or finding objects within defined boundaries. One main concern is to restrict the search space. Different solutions to this problem have been proposed. Presented in this paper is a double dynamic programming approach which is both optimal and computationally fast. It employs dynamic programming for finding correspondence between pairs of pixels on the inner and outer boundary of a band, which is found through morphological transforms and restricts the search space. In a second step dynamic programming is used again to find the exact contour inside this restricted search space.

• Type: Conference
• Year: 2004
• Where: ICIAR
• Authors: Mohammad Dawood, Xiaoyi Jiang, Klaus P. Schäfers
• Keywords: Constrained Search Space, Dynamic Programming, ICIAR 2004, Search Space