Share Email Print

Proceedings Paper

A weak and small target detection algorithm based on TMS320C6678 for convolutional neural network
Author(s): Zhuo He; Jin Lu; Chen Xu; Tianzhang Nan
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

TMS320C6678 is a multi-core DSP processor with higher processing speed compared with other single-core DSPs. This paper proposes a convolutional neural system based on the technical requirements of TMS320C6678 processing target detection technology to enhance real-time and reduce false alarm rate. Infrared weak target detection algorithm for networks. Pre-processing methods such as image denoising and enhancement and Kronecker product up sampling are used to expand the target size and enhance the morphological features while reducing the difficulty of detecting weak targets. Then the design is more suitable for detecting the convolutional neural network structure of weak targets. Extracting the deep features of infrared weak targets can accurately detect weak targets in infrared images while effectively eliminating various disturbances such as features and clouds. The simulation results show that the target detection probability of the algorithm is 95%, with low time complexity, strong resistance to ground and cloud interference, and high detection accuracy. At present, the core part of the algorithm is transplanted into TMS320C6678, the processing speed is much higher than the simulation result, the real-time performance is confirmed, and it has high engineering application value.

Paper Details

Date Published: 31 January 2020
PDF: 6 pages
Proc. SPIE 11427, Second Target Recognition and Artificial Intelligence Summit Forum, 114273D (31 January 2020);
Show Author Affiliations
Zhuo He, Tianjin Jinhang Institute of Technical Physics (China)
Jin Lu, Tianjin Jinhang Institute of Technical Physics (China)
Chen Xu, Tianjin Jinhang Institute of Technical Physics (China)
Tianzhang Nan, Tianjin Jinhang Institute of Technical Physics (China)

Published in SPIE Proceedings Vol. 11427:
Second Target Recognition and Artificial Intelligence Summit Forum
Tianran Wang; Tianyou Chai; Huitao Fan; Qifeng Yu, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?