Spectra of shape contexts: An application to symbol recognition

作者：

Highlights：

• Pixel-level constraint (PLC) histogram is robust but with O(N3) complexity.

• Spectra of shape contexts (SSC) inherit the robustness with reduced complexity.

• Through rigorous theoretical deduction, PLC is correlated to SSC via FFT.

• SSC outperforms PLC in robustness and shape context in invariance experimentally.

摘要

•Pixel-level constraint (PLC) histogram is robust but with O(N3) complexity.•Spectra of shape contexts (SSC) inherit the robustness with reduced complexity.•Through rigorous theoretical deduction, PLC is correlated to SSC via FFT.•SSC outperforms PLC in robustness and shape context in invariance experimentally.

论文关键词：Symbol recognition,Shape matching,Histogram-based descriptor,Shape context,FFT,Histogram comparison,Point pattern matching,Maximum clique,Image retrieval,Object recognition

论文评审过程：Received 30 September 2012, Revised 26 October 2013, Accepted 18 November 2013, Available online 1 December 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.11.020