一种用于图像翘曲变换的高速缓冲存储器设计

康洁; 刘强

doi:10.13700/j.bh.1001-5965.2019.0284

一种用于图像翘曲变换的高速缓冲存储器设计

doi: 10.13700/j.bh.1001-5965.2019.0284

康洁^{1, 2},
刘强^{1, 2, ,}

1.
天津大学微电子学院, 天津 30007
2.
天津市成像与感知微电子技术重点实验室, 天津 300072

基金项目:

国家自然科学基金 61574099

详细信息

作者简介:
康洁女, 硕士研究生。主要研究方向:用于高效平视显示系统的高速缓冲存储器设计

刘强男, 博士, 副教授, 博士生导师。主要研究方向:数字集成电路设计、低功耗电路设计等

通讯作者:
刘强, E-mail: qiangliu@tju.edu.cn

中图分类号: TN402
计量
- 文章访问数: 688
- HTML全文浏览量: 135
- PDF下载量: 136
- 被引次数: 0
出版历程
- 收稿日期: 2019-06-10
- 录用日期: 2019-09-29
- 网络出版日期: 2020-04-20

A Cache design for image warping

KANG Jie^{1, 2},
LIU Qiang^{1, 2
, ,}

1.
School of Microelectronics, Tianjin University, Tianjin 30007
2.
Tianjin Key Laboratory of Imaging and Sensing Microelectronic Technology, Tianjin 300072, China

Funds:

National Natural Science Foundation of China 61574099

More Information

Corresponding author: LIU Qiang, E-mail: qiangliu@tju.edu.cn

摘要

摘要:
车载平视显示（HUD）系统通过图像翘曲变换将原始平面图像信息显示在挡风玻璃曲面上，原始图像数据的非线性访问会造成存储器访问效率下降。为此，设计了一种高速缓冲存储器（Cache），以最大程度保证像素数据访问的连续性，减少存储器访问次数并提高带宽资源利用率。为优化Cache性能，提出存储空间分离管理技术和地址分级比较技术，提高图像像素在Cache中的存储密度，并节省逻辑资源。此外，提出一种Cache容量动态调整的方法，在保证命中率前提下减少Cache存储资源的使用、降低功耗。实验结果显示，存储空间分离管理技术使存储资源节省25%，地址分级比较技术使逻辑资源节省近10%，Cache容量可以减少75%，且动态功耗减少67.578%，静态功耗减少14.060%。
- 图像翘曲变换 /
- 高速缓冲存储器 /
- 存储空间分离管理 /
- 地址分级比较 /
- 容量动态调整
Abstract:
The vehicle head-up display (HUD) system displays the source image plane on the windshield surface by image warping. Non-linear access to source image pixels degrades memory access efficiency. To solve this problem, a Cache is designed to ensure pixel access continuity, reduce memory access numbers and improve bandwidth resource utilization. To optimize the performance of Cache, a storage separation management technology and an address multi-level comparison technology are proposed to improve the storage density of image pixels and save logical resources. In addition, a method for dynamically adjusting the Cache capacity is proposed to reduce storage resource usage and power consumption while ensuring the hit rate. The experimental results show that the storage separation management technology saves storage resources by 25%. The address multi-level comparison technology saves logical resources by nearly 10%. The Cache capacity can be reduced by the dynamic adjusting method by 75%, dynamic power consumption is reduced by 67.578%, and static power consumption is reduced by 14.060%.
- image warping /
- Cache /
- storage separation management /
- address multi-level comparison /
- capacity dynamic adjustment

HTML全文

图 1 图像翘曲变换像素坐标映射关系

Figure 1. Coordinate mapping of pixels in image warping

下载: 全尺寸图片幻灯片

图 2 Cache块存储结构

Figure 2. Storage structure of Cache block

下载: 全尺寸图片幻灯片

图 3 存储空间分离管理技术示意图

Figure 3. Schematic diagram of storage separation management technology

下载: 全尺寸图片幻灯片

图 4 Cache地址分级比较过程

Figure 4. Multi-level comparison process of Cache addresses

下载: 全尺寸图片幻灯片

图 5 LRU替换算法示意图

Figure 5. Schematic diagram of LRU replacement algorithm

下载: 全尺寸图片幻灯片

表 1 Cache容量动态调整技术对实验结果的影响

Table 1. Effect of Cache capacity dynamic adjustment technology on experimental results

Cache块数	命中率/%	动态功耗/W	静态功耗/W
64	98.786	20.076	3.293
32	98.736	11.610	2.988
16	98.716	6.509	2.830
变化量/%	-0.071	-67.578	-14.060

下载: 导出CSV

表 2 存储空间分离管理技术对资源使用情况的影响

Table 2. Effect of storage separation management technology on resource usage

资源	Cache块数	改进前	改进后	变化量/%
CLB LUTs	64	5182	6594	27.248
	32	2628	2951	12.291
	16	1209	1387	14.723
LUTRAM	64	2048	1536	-25.000
	32	1024	768	-25.000
	16	512	384	-25.000
CLB Registers	64	3743	3242	-13.385
	32	1855	1604	-13.531
	16	927	807	-12.945

下载: 导出CSV

表 3 地址分级比较技术对资源使用情况的影响

Table 3. Effect of address multi-level comparison technology on resource usage

资源	Cache块数	改进前	改进后	变化量/%
CLB LUTs	64	5182	5252	1.351
	32	2628	2559	-2.626
	16	1209	1217	0.662
CLB Registers	64	3743	3374	-9.858
	32	1855	1710	-7.817
	16	927	842	-9.169

下载: 导出CSV

参考文献(17)

[1]	LEE M, KIM H, PAIK J.Correction of barrel distortion in fisheye lens images using image-based estimation of distortion parameters[J].IEEE Access, 2019, 7:45723-45733. doi: 10.1109/ACCESS.2019.2908451
[2]	YUAN X, HU C, CHEN J, et al.Correction of capsule endoscope image distortion based on two-dimensional look-up table[C]//Proceeding of the 11th World Congress on Intelligent Control and Automation.Piscataway, NJ: IEEE Press, 2014: 553-557.
[3]	SHETE P P, MADHUKAR S D, BOSE S K.A real-time stereo rectification of high definition image stream using GPU[C]//2014 International Conference on Advances in Computing, Communications and Informatics (ICACCI).Piscataway, NJ: IEEE Press, 2014: 158-162.
[4]	MENG G, XIANG S, PAN C, et al.Active rectification of curved document images using structured beams[J].International Journal of Computer Vision, 2017, 122(1):34-60.
[5]	TAN Z, ZHANG S, WANG R.Stable stitching method for stereoscopic panoramic video[J].CAAI Transactions on Intelligence Technology, 2018, 3(1):1-7.
[6]	LIN C, TSAI Y M, WANG W, et al.GPU-accelerated high-resolution image stitching with better initial guess[C]//2018 IEEE International Conference on Consumer Electronics(ICCE).Piscataway, NJ: IEEE Press, 2018: 1-3.
[7]	MELO R, FALCAO G, BARRETO J P.Real-time HD image distortion correction in heterogeneous parallel computing systems using efficient memory access patterns[J].Journal of Real-Time Image Processing, 2016, 11(1):83-91. doi: 10.1007/s11554-012-0304-3
[8]	RYOO J R, LEE E S, PARK H K.Real-time implementation of an LUT-based image warping system[C]//IEEE Intelligence and Safty for Robotics 2013.Piscataway, NJ: IEEE Press, 2013: 1-4.
[9]	RYOO J R, LEE E S, DOH T Y.An implementation of real-time image warping using FPGA[J].Journal of Embedded Systems & Applications, 2014, 9(6):335-344.
[10]	LU Y, LUO X, WANG Y, et al.Line buffer reduction for LUT-based real-time image inverse warping[C]//2016 14th IEEE International New Circuits and Systems Conference(NEWCAS).Piscataway, NJ: IEEE Press, 2016: 1-4.
[11]	GREISEN P, HEINZLE S, GROSS M, et al.An FPGA-based processing pipeline for high-definition stereo video[J].EURASIP Journal on Image & Video Processing, 2011, 2011:18.
[12]	JOON C Y, RAE R J.Image Cache for FPGA-based real-time image warping[J].Journal of the Institute of Electronics and Information Engineers, 2016, 53(6):91-100. doi: 10.5573/ieie.2016.53.6.091
[13]	HAN X, CHEN S L, WU L, et al.Design and verification of distributed RAM using look-up tables in an SOI-based FPGA[C]//2010 10th IEEE International Conference on Solid-State and Integrated Circuit Technology.Piscataway, NJ: IEEE Press, 2010: 306-308.
[14]	PENG M, LIU X.Adaptive rapid reconfigurable algorithm for low power Cache[C]//International Conference on Computational & Information Sciences.Piscataway, NJ: IEEE Press, 2013: 203-206.
[15]	UPADHYAY B R, SUDARSHAN T S B.Low power predictive placement Cache scheme for embedded system[C]//2014 International Conference on Embedded Systems(ICES).Piscataway, NJ: IEEE Press, 2014: 250-254.
[16]	KUMAR S, SINGH P K.An overview of modern Cache memory and performance analysis of replacement policies[C]//2016 IEEE International Conference on Engineering and Technology(ICETECH).Piscataway, NJ: IEEE Press, 2016: 210-214.
[17]	杜建海, 吕江花, 高世伟, 等.面向航天器综合测试系统的Web缓存替换策略[J].北京航空航天大学学报, 2018, 44(8):1609-1619. doi: 10.13700/j.bh.1001-5965.2017.0591 DU J H, LYU J H, GAO S W, et al.A Web Cache replacement strategy for spacecraft comprehensive testing system[J].Journal of Beijing University of Aeronautics and Astronautics, 2018, 44(8):1609-1619(in Chinese). doi: 10.13700/j.bh.1001-5965.2017.0591