Qiwei Jiang
Research Subjects
Qiwei Jiang has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Wei Xiong
College of Computer Science, National Key Laboratory of Fundamental Science on Synthetic Vision, Sichuan University, Chengdu, Sichuan, China
1 shared publication - Xing Zhang
College of Electrical Engineering, Sichuan University, Chengdu, Sichuan, CHINA
1 shared publication - Yanqiong Guo
College of Computer Science, National Key Laboratory of Fundamental Science on Synthetic Vision, Sichuan University, Chengdu, Sichuan, China
1 shared publication
Latest Publications
- 27 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
An Efficient Pipeline Architecture Based on Spatio-Temporal Stereo Matching for Free-Pose 3-D Facial Measurement
Implementing high-precision, real-time stereo matching with large disparity levels on FPGAs is essential for free-pose 3D facial measurement. Current solutions are mainly limited by the constrained accuracy of single texture image processing and the significant computational cost of the expansion of disparity level. Leveraging the accuracy gains of Spatio-Temporal (ST) stereo-matching, this paper implements an efficient FPGA-based pipelined architecture tailored for ST stereo matching on a speckle-structured-light 3D measurement system, thereby overcoming these challenges. The architecture segments the disparity search range into multiple partitions, enabling the parallel processing of 8 pixel disparities per clock cycle through a balanced resource allocation. By integrating the Generalized Census Transform (GCT) and a 5×5 matching window, we achieve a disparity throughput of 1024×800 @ 55.5 FPS with a disparity level of 512 and a power consumption of 6.15 W, on a Zynq XC7Z100 platform operating at 200 MHz, and take 167.2k LUTs, 159.8k FFs, 103.5 BRAMs, and 304 DSPs. Experimental results verify that our proposed architecture achieves a good balance between accuracy, real-time performance, disparity levels, and resource costs, offering broader implications for real-time, high-precision 3D facial expression measurement in free poses.
- 27 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
An Efficient Pipeline Architecture Based on Spatio-Temporal Stereo Matching for Free-Pose 3-D Facial Measurement
Implementing high-precision, real-time stereo matching with large disparity levels on FPGAs is essential for free-pose 3D facial measurement. Current solutions are mainly limited by the constrained accuracy of single texture image processing and the significant computational cost of the expansion of disparity level. Leveraging the accuracy gains of Spatio-Temporal (ST) stereo-matching, this paper implements an efficient FPGA-based pipelined architecture tailored for ST stereo matching on a speckle-structured-light 3D measurement system, thereby overcoming these challenges. The architecture segments the disparity search range into multiple partitions, enabling the parallel processing of 8 pixel disparities per clock cycle through a balanced resource allocation. By integrating the Generalized Census Transform (GCT) and a 5×5 matching window, we achieve a disparity throughput of 1024×800 @ 55.5 FPS with a disparity level of 512 and a power consumption of 6.15 W, on a Zynq XC7Z100 platform operating at 200 MHz, and take 167.2k LUTs, 159.8k FFs, 103.5 BRAMs, and 304 DSPs. Experimental results verify that our proposed architecture achieves a good balance between accuracy, real-time performance, disparity levels, and resource costs, offering broader implications for real-time, high-precision 3D facial expression measurement in free poses.