The Effect of Pose on the distribution of Edge Gradients in Omnidirectional Images

Jarvis, Dean; Kyriacou, Theocharis

Abstract

Images from omnidirectional cameras are used frequently in applications involving artificial intelligence and robotics as a source of rich information about the surroundings. A useful feature that can be extracted from these images is the distribution of gradients of the edges in the scene. This distribution is affected by the pose of the camera on-board a robot at any given location in the environment. This paper investigates the effect of the pose on this distribution. The gradients in the images are extracted and arranged into a histogram which is then compared to the histograms of other images using a chi-squared test. It is found that any differences in the distribution are not specific to either the position or orientation and that there is a significant difference in the distributions of two separate locations. This can aid in the localisation of robots when navigating.

Citation

Jarvis, D., & Kyriacou, T. (2018, July). The Effect of Pose on the distribution of Edge Gradients in Omnidirectional Images. Presented at 19th Annual Conference, TAROS 2018, Bristol, UK

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 19th Annual Conference, TAROS 2018
• Start Date: Jul 25, 2018
• End Date: Jul 27, 2018
• Acceptance Date: Apr 12, 2018
• Publication Date: Jul 21, 2018
• Publisher: Springer
• Series Title: Towards Autonomous RObotic Systems (TAROS)
• Book Title: Towards Autonomous Robotic Systems. TAROS 2018.
• ISBN: 9783319967288
• DOI: https://doi.org/10.1007/978-3-319-96728-8_20
• Keywords: Jarvis D., Kyriacou T. (2018) The Effect of Pose on the Distribution of Edge Gradients in Omnidirectional Images. In: Giuliani M., Assaf T., Giannaccini M. (eds) Towards Autonomous Robotic Systems. TAROS 2018. Lecture Notes in Computer Science, vol 10965. Springer, Cham
• Public URL: https://keele-repository.worktribe.com/output/410885
• Publisher URL: https://link.springer.com/chapter/10.1007/978-3-319-96728-8_20