Genetic Programming for Extracting Edge Features Using Two Blocks

Fu, W., Zhang, M. and Johnston, M.

    In low-level edge detection, single pixels have been popularly used to extract edge features. However, the extracted edge features might not have good abil- ity to effectively mark edge points on images with noise or/and textures. Single pixels can be extracted based on a local window. To automatically search pix- els to extract edge features using Genetic Program- ming, search operators based on single pixels and sin- gle blocks of pixels have been proposed. Single blocks of pixels can be used to improve detection perfor- mance on natural images, but the computational cost is high. In this paper, to reduce the computational cost of using blocks of pixels, a new search opera- tor based on two blocks of pixels is proposed. The experiment results show that the proposed search op- erator can effectively reduce computational cost on evolved edge detectors, remaining good detection per- formance.
Cite as: Fu, W., Zhang, M. and Johnston, M. (2015). Genetic Programming for Extracting Edge Features Using Two Blocks. In Proc. Thirteenth Australasian Data Mining Conference (AusDM 2015) Sydney, Australia. CRPIT, 168. Ong, K.L., Zhao, Y., Stone, M.G. and Islam, M.Z. Eds., ACS. 141-150
pdf (from crpit.com) pdf (local if available) BibTeX EndNote GS