Previous [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

Journal of Inforamtion Science and Engineering, Vol.17 No.1, pp.133-145 (January 2001)

Modified Hough Transforms for Object
Feature Extraction*

Jar-Ferr Yang and Shu-Sheng Hao+
Department of Electrical Engineering
National Cheng Kung University
Tainan, Taiwan 701, R.O.C.

In this paper, we propose the use of modified Hough transforms to efficiently extract object feature parameters, which are usually contaminated by heavily noisy corrugation and discontinuity. The modified HT (MHT) is developed by introducing spatial and parameter weighting functions to imporve the detection performance for the traditional Hough transform (HT), which generally fails to robustly detect natural object parameters. Using designed test patterns and real images, simulations show that the proposed weighting functions are helpful in detecting noise-corrupted objec features. Due to its robustness, the MHT can be easily figured with a coarse-to-fine adaptive search mechanism to reduce the huge amount of computation for feature parameters extraction.

Keywords: modified Hough transform, model-based coding, feature parameters extraction, coarse-to-fine search, facial object estimation

Full Text () Retrieve PDF document (200101_09.pdf : 3,563,874 bytes)

Received July 2, 1999; revised November 16, 1999; accepted December 30, 1999.
Communicated by Mark H. Y. Liao.
*This research was supported by National Science Council under Contact #NSC-85-2221-E-006-046, Taiwan, R.O.C.