100 lines of code for shape-based object localization

作者：

Highlights：

• New segment-based representation that we call OPS.

• Used to achieve high flexibility that is required to account for intra-class variations.

• OPS representation is fed to a dynamic programming algorithm for robust detection.

• Very competitive localization results under extensive noise and clutter.

• The ability to accurately localize detailed object boundaries.

摘要

Highlights•New segment-based representation that we call OPS.•Used to achieve high flexibility that is required to account for intra-class variations.•OPS representation is fed to a dynamic programming algorithm for robust detection.•Very competitive localization results under extensive noise and clutter.•The ability to accurately localize detailed object boundaries.

论文关键词：Object detection,Boundary localization,Deformable templates,Dynamic programming,Contour segments,One-pixel segment (OPS)

论文评审过程：Received 6 April 2016, Revised 2 June 2016, Accepted 4 June 2016, Available online 7 June 2016, Version of Record 20 June 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.06.003