Image feature point extraction based on neural network  is implemented on FPGA

Qinglin Fu; Jun Li

doi:10.1117/12.2540140

14 August 2019 Image feature point extraction based on neural network is implemented on FPGA

Qinglin Fu, Jun Li

Proceedings Volume 11179, Eleventh International Conference on Digital Image Processing (ICDIP 2019); 1117947 (2019) https://doi.org/10.1117/12.2540140
Event: Eleventh International Conference on Digital Image Processing (ICDIP 2019), 2019, Guangzhou, China

Abstract

The traditional image feature point extraction relies on the characteristics of manual design. This paper proposes a neural network image feature point extraction method based on FPGA, which detects image feature points, direction estimation and descriptor extraction. Each part is Based on the convolutional neural network (CNN) implementation, using pipeline optimization, loop unrolling, storage optimization, fixed-point quantization, etc., using Xilinx's high-level synthesis tool Vivado HLS, the algorithm programs of the neural network layers written by C++ and OpenCV are converted into At the RTL level, the image feature points are extracted using the "Python+ARM+FPGA" method. Experiments show that FPGA can extract a large number of image feature points, and FPGA is 8.2 times more efficient than CPU, 1.7 times that of GPU, and power consumption is much lower than GPU.

Citation Download Citation

Qinglin Fu and Jun Li "Image feature point extraction based on neural network is implemented on FPGA", Proc. SPIE 11179, Eleventh International Conference on Digital Image Processing (ICDIP 2019), 1117947 (14 August 2019); https://doi.org/10.1117/12.2540140

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available