WCSE

Polar Cosine-Sine Transform for Image Representation

Abstract— Since Polar Harmonic Transforms (PHTs) have been introduced, they are widely used in image analysis and pattern recognition with computation of the kernel is exceedingly briefness. However, PHTs always have weakness of instability at high orders of moments. What's worse? PCET encounters this undesirable situation even at low orders. This paper presents the Polar Cosine-Sine Transform (PCST) using a new radial kernel based on Cosine-Sine functions to achieve all-rights stability. First, we analyze the radial kernel and the reason of instability of the traditional PCET. Second, we structure a new radial kernel using Cosine-Sine functions based on analysis of PCET. Finally, PCST has been introduced. The experimental result shows that the proposed method holds rotation invariance and the stability of PCST is superior to PHTs in terms of image representation capability when high-order moments are concerned.

Index Terms— Orthogonal moments, Polar harmonic transforms, Rotation invariants, Polar cosine-sine transform.

Kezheng Sun, Lijuan Tang
School of Info., Vocational Col. of Bus., CHINA
Lijuan Tang
School of Info. and Elec. Eng., China Univ. of Mining and Technology, CHINA

[Download]

Cite: Kezheng Sun, Lijuan Tang, "Polar Cosine-Sine Transform for Image Representation," Proceedings of 2016 6th International Workshop on Computer Science and Engineering, pp. 243-248, Tokyo, 17-19 June, 2016.