Feature Detection Performance Based Benchmarking of Motion Deblurring Methods: Applications to Vision for Legged Robots

作者：

Highlights：

• Demonstrated unsuitability of available image quality metrics in terms of image feature extraction and computer vision.

• A novel feature detection based performance metric (FD-AROC) is proposed.

• First multi-sensor motion blur data set on a hexapedal legged robot with ground truth motion data.

• Comparison of deblurring methods using the proposed data set and the FD-AROC metric.

摘要

•Demonstrated unsuitability of available image quality metrics in terms of image feature extraction and computer vision.•A novel feature detection based performance metric (FD-AROC) is proposed.•First multi-sensor motion blur data set on a hexapedal legged robot with ground truth motion data.•Comparison of deblurring methods using the proposed data set and the FD-AROC metric.

论文关键词：Feature detection,Computer vision,Motion blur,Motion deblurring,Blur metric,Legged locomotion,Robotics,Computer vision dataset

论文评审过程：Received 24 July 2017, Revised 13 October 2018, Accepted 9 January 2019, Available online 10 February 2019, Version of Record 13 March 2019.

论文官网地址：https://doi.org/10.1016/j.imavis.2019.01.002