基于嵌入式GPU的运动目标分割算法并行优化

在光电监视系统中, 广泛应用于运动目标分割的PBAS(pixel base adaptive segmenter)算法计算复杂、参数量大, 难以达到实时分割的要求。针对PBAS算法是对图像中每个像素点进行独立处理, 特别适合于GPU并行加速的特点, 对其在嵌入式GPU平台Jetson TX2上进行了并行优化实现。在数据存储结构、共享内存使用、随机数产生机制3个方面对该算法进行了优化设计。实验结果表明, 对于480×320像素分辨率的中波红外视频序列, 该并行优化方法可以达到132 fps的处理速度, 满足了实时处理的要求。

Abstract

In optoelectronic surveillance systems, the pixel base adaptive segmenter (PBAS) algorithm, which is widely used in moving objects segmentation, is hard to meet the requirements of real-time applications due to its calculating complication and a large amount of computing parameters. With its pixel-level parallelism, deploying PBAS on top of graphic processing unit (GPU) is promising. This paper implements real-time optimization of PBAS on embedded GPU platform-Jetson TX2, employing methods of data storage architecture, shared memory utilization and random number generation. Experimental results show that the parallel optimization method can achieve 132 fps when processing 480×320 pixel medium-wave infrared video sequences, thus meets the real-time processing need.

参考文献

[1] STAUFFER C, GRIMSON W E L. Adaptive background mixture models for real-time tracking［J］.Computer Society Conference on Computer Vision and Pattern Recognition,1999,2: 2246.

[2] HOFMANN M, TIEFENBACHER P, RIGOLL G. Background segmentation with feedback: the pixel-based adaptive segmenter［C］.USA: IEEE, 2012.

[3] BARNICH O, VAN DROOGENBROECK M. ViBe: A universal background subtraction algorithm for video sequences［J］. IEEE Transactions on Image Processing, 2011, 20(6): 1709-1724.

[4] NVIDIA.CUDA C Programming Guide,PG-02829-001_v8.0, 2017.5 ［EB/OL］.［2017-10-15］. https: //docs.nvidia.com/cuda/cuda-c-programming-guide.

[5] NVIDIA.CUDA C Best Practices Guide,DG-05603-001_v8.0, 2017.5 ［EB/OL］.［2017-10-15］. https: //docs.nvidia.com/cuda/cuda-c-best-practices-guide.

[6] NVIDIA. CUDA Quick Start Guide, DU-05347-301_v8.0, 2017.5 ［EB/OL］. ［2018-09-10］.https: //docs.nvidia.com/cuda/cuda-quick-start-guide.

[7] NVIDIA.NVIDIA Jetson TX2 System-on-Modul, v1.1, 20e［EB/OL］. ［2017-12-12］.https: //developer.nvidia.com/embedded/jetson-tx2.

[8] NVIDIA.Jetson TX1 and TX2 Developer Kits,DA_07976_001_01, 2017.7［EB/OL］. ［2017-12-10］.https: //developer.nvidia.com/embedded/dlc/l4t-28-1-jetson-developer-kit-user-guide.

[9] NVIDIA. Pascal Architecture Whitepaper,V1.2, 2017［EB/OL］. ［2017-12-13］.https: //images.nvidia.com/content/pdf/tesla/whitepaper/pascal-architecture-whitepaper.pdf.

[10] 左颢睿, 张启衡, 徐勇, 等. 基于GPU的快速Sobel边缘检测算法［J］. 光电工程, 2009, 36(1): 8-12.

ZUO Haorui, ZHANG Qiheng, XU Yong, et al. Fast sobel edge detection algorithm based on GPU［J］. Opto-Electronic Engineering, 2009, 36(1): 8-12.

[11] 唐滔, 彭林, 黄春, 等. 面向存储层次设计优化的GPU程序性能分析［J］. 计算机科学, 2017, 44(12): 1-10.

TANG Tao, PENG Lin, HUANG Chun, et al. Performance analysis of GPU programs towards better memory hierarchy design［J］. Computer Science, 2017, 44(12): 1-10.

[12] 严发宝, 苏艳蕊, 赵占锋, 等. 基于GPU的小尺寸FFT在实时图像复原中的优化［J］. 中南大学学报: 自然科学版, 2017, 48(10): 2691-2696.

YAN Fabao, SU Yanrui, ZHAO Zhanfeng, et al. Optimization on FFT of small size in real-time image restoration based on GPU［J］. Journal of Central South University: Science and Technology, 2017, 48(10): 2691-2696.

[13] PARK S I, PONCE S P, HUANG J, et al. Low-cost, high-speed computer vision using NVIDIAs CUDA architecture［C］.USA: IEEE, 2008.

[14] BERNABE S, LOPEZ S, PLAZA A, et al. GPU implementation of an automatic target detection and classification algorithm for hyperspectral image analysis［J］. IEEE Geoscience and Remote Sensing Letters, 2013, 10(2): 221-225.

[15] YAO G L, LEI T, ZHONG J D, et al. Comparative evaluation of background subtraction algorithms in remote scene videos captured by MWIR sensors［J］. Sensors, 2017, 17(9): 1945.

[16] MARTIN P J, AYUSO L F, TORRES R, et al. Algorithmic strategies for optimizing the parallel reduction primitive in CUDA［C］.USA: IEEE, 2012.

[17] ELLIOTT G A, ANDERSON J H. Real-world constraints of GPUs in real-time systems［C］. USA: IEEE, 2011.

[18] NVIDIA.Curand Library Programming Guide, PG-05328-050 _v8.0, 2016.2［EB/OL］.［2018-12-15］. https: //developer.nvidia.com/curand.

张刚, 马震环, 雷涛, 崔毅, 张三喜. 基于嵌入式GPU的运动目标分割算法并行优化[J]. 应用光学, 2019, 40(6): 1067. ZHANG Gang, MA Zhenhuan, LEI Tao, CUI Yi, ZHANG Sanxi. Embedded GPU-based parallel optimization for moving objects segmentation algorithm[J]. Journal of Applied Optics, 2019, 40(6): 1067.

基于嵌入式GPU的运动目标分割算法并行优化

关于本站 Cookie 的使用提示

全站搜索

基于嵌入式GPU的运动目标分割算法并行优化

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索