AI 2012 - LNAI 7691

Automatic Construction of Invariant Features Using Genetic Programming for Edge Detection

Wenlong Fu¹, Mark Johnston¹, and Mengjie Zhang²

¹School of Mathematics, Statistics and Operations Research, Victoria University of Wellington, PO Box 600, Wellington, New Zealand

²School of Engineering and Computer Science, Victoria University of Wellington, PO Box 600, Wellington, New Zealand

Abstract. This paper investigates automatic construction of invariant features using Genetic Programming (GP) for edge detection. Generally, basic features for edge detection, such as gradients, are further manipulated to improve detection performance. In order to improve detection performance, new features are constructed from different local features. In this study, GP is proposed to automatically construct invariant features based on basic invariant features from gradients, image quality (means and standard deviations), and histograms of images. The experimental results show that the invariant features constructed by GP combine advantages from the basic features, reduce drawbacks from basic features alone, and also improve the detection performance.

Keywords: Genetic Programming, Edge Detection, Image Analysis, Feature Construction

LNAI 7691, p. 144 ff.

Full article in PDF | BibTeX