Stream-based ORB feature extractor with dynamic power optimization
The Oriented Fast and Rotated BRIEF (ORB) feature extractor, which consists of key-point detection and descriptor computation, is a key module in many computer vision systems. Existing hardware implementations of ORB feature extractor only focus on increasing performance with power optimization as a...
| Main Authors: | , , , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/147464 |
| _version_ | 1826123842235400192 |
|---|---|
| author | Tran, Phong Pham, Thinh Hung Lam, Siew-Kei Wu, Meiqing Jasani, Bhavan A. |
| author2 | School of Computer Science and Engineering |
| author_facet | School of Computer Science and Engineering Tran, Phong Pham, Thinh Hung Lam, Siew-Kei Wu, Meiqing Jasani, Bhavan A. |
| author_sort | Tran, Phong |
| collection | NTU |
| description | The Oriented Fast and Rotated BRIEF (ORB) feature extractor, which consists of key-point detection and descriptor computation, is a key module in many computer vision systems. Existing hardware implementations of ORB feature extractor only focus on increasing performance with power optimization as a post consideration. In this paper, we present a stream-based ORB feature extractor that incorporates mechanisms to lower the dynamic power consumption. These mechanisms exploit the fact that the number of detected keypoints is typically small. The proposed solution significantly lowers the switching activity of the key-point detection and descriptor computation stages by early pruning of non-likely key-points and gating the descriptor computation stages. Further power reduction and resource minimization are achieved by employing a threshold-guided bit-width optimization strategy to truncate the redundant bits in the key-point detection stage. Finally, we propose an approximation method to achieve rotation invariance of the descriptors. FPGA implementation targeting the Altera Aria V device shows that the proposed strategies lead to over 25% reduction in dynamic power and lower resource utilization, with only marginal loss in accuracy. |
| first_indexed | 2024-10-01T06:10:56Z |
| format | Conference Paper |
| id | ntu-10356/147464 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T06:10:56Z |
| publishDate | 2021 |
| record_format | dspace |
| spelling | ntu-10356/1474642021-04-09T04:59:07Z Stream-based ORB feature extractor with dynamic power optimization Tran, Phong Pham, Thinh Hung Lam, Siew-Kei Wu, Meiqing Jasani, Bhavan A. School of Computer Science and Engineering Proceedings of the 2018 International Conference on Field-Programmable Technology (FPT) Engineering::Computer science and engineering::Hardware Feature Extraction Hardware Acceleration The Oriented Fast and Rotated BRIEF (ORB) feature extractor, which consists of key-point detection and descriptor computation, is a key module in many computer vision systems. Existing hardware implementations of ORB feature extractor only focus on increasing performance with power optimization as a post consideration. In this paper, we present a stream-based ORB feature extractor that incorporates mechanisms to lower the dynamic power consumption. These mechanisms exploit the fact that the number of detected keypoints is typically small. The proposed solution significantly lowers the switching activity of the key-point detection and descriptor computation stages by early pruning of non-likely key-points and gating the descriptor computation stages. Further power reduction and resource minimization are achieved by employing a threshold-guided bit-width optimization strategy to truncate the redundant bits in the key-point detection stage. Finally, we propose an approximation method to achieve rotation invariance of the descriptors. FPGA implementation targeting the Altera Aria V device shows that the proposed strategies lead to over 25% reduction in dynamic power and lower resource utilization, with only marginal loss in accuracy. National Research Foundation (NRF) This research project is partially funded by the National Research Foundation Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. 2021-04-09T04:55:39Z 2021-04-09T04:55:39Z 2018 Conference Paper Tran, P., Pham, T. H., Lam, S., Wu, M. & Jasani, B. A. (2018). Stream-based ORB feature extractor with dynamic power optimization. Proceedings of the 2018 International Conference on Field-Programmable Technology (FPT), 97-104. https://dx.doi.org/10.1109/FPT.2018.00024 9781728102139 https://hdl.handle.net/10356/147464 10.1109/FPT.2018.00024 2-s2.0-85068324957 97 104 en © 2018 Institute of Electrical and Electronics Engineers (IEEE). All rights reserved. |
| spellingShingle | Engineering::Computer science and engineering::Hardware Feature Extraction Hardware Acceleration Tran, Phong Pham, Thinh Hung Lam, Siew-Kei Wu, Meiqing Jasani, Bhavan A. Stream-based ORB feature extractor with dynamic power optimization |
| title | Stream-based ORB feature extractor with dynamic power optimization |
| title_full | Stream-based ORB feature extractor with dynamic power optimization |
| title_fullStr | Stream-based ORB feature extractor with dynamic power optimization |
| title_full_unstemmed | Stream-based ORB feature extractor with dynamic power optimization |
| title_short | Stream-based ORB feature extractor with dynamic power optimization |
| title_sort | stream based orb feature extractor with dynamic power optimization |
| topic | Engineering::Computer science and engineering::Hardware Feature Extraction Hardware Acceleration |
| url | https://hdl.handle.net/10356/147464 |
| work_keys_str_mv | AT tranphong streambasedorbfeatureextractorwithdynamicpoweroptimization AT phamthinhhung streambasedorbfeatureextractorwithdynamicpoweroptimization AT lamsiewkei streambasedorbfeatureextractorwithdynamicpoweroptimization AT wumeiqing streambasedorbfeatureextractorwithdynamicpoweroptimization AT jasanibhavana streambasedorbfeatureextractorwithdynamicpoweroptimization |