Integrated adaptive particle filtering-based color digital image correlation for large deformation measurement
Year: 2024
Type: article
Abstract: In large deformation measurement, digital image correlation (DIC[1]) using functional forms can no longer describe the complex changes in grayscale information and morphology of sub-regions. To address this issue, particle filtering (PF) can be combined with color DIC (PF-DIC[2]) to establish a color distribution model to describe sub-regions, but traditional PF has poor performance. Therefore, this paper proposes an integrated adaptive particle filtering-based color digital image correlation (APF-DIC[3]). This method first constructs a dynamic color distribution model based on ACFEF[4] to describe the sub-region. It then introduces the SSKL[5] correlation coefficient to measure the similarity between sub-regions, and establishes TAR[6] to adaptively adjust the number of particles, ultimately achieving adaptive matching of sub-regions under complex deformations. Performance evaluation and simulation results show that APF-DIC significantly improves the accuracy, robustness, and computational efficiency of the algorithm. Real experimental results further verify the effectiveness of APF-DIC, demonstrating its excellent illumination invariance. [1] Digital image correlation (DIC).[2] Particle filtering (PF) can be combined with color DIC, that is, particle filter-based color digital image correlation (PF-DIC).[3] Integrated adaptive particle filtering-based color digital image correlation (APF-DIC).[4] Adaptive color feature extraction function (ACFEF).[5] Smoothed symmetric Kullback Leibler (SSKL).[6] Hyperbolic tangent adaptive resampling (tanh-adaptive resampling, TAR).
Source: Optics & Laser Technology
Authors: Xiao-Yong Liu, Xiao-Wei Zhang, Rong-Li Li, ZhaoPeng Hao, Kai-Kai Li, Xiao-Ri Pei, Dongwei Gu, Qihan Li, Jia-Ming Hu, Guo-Qing Han
Institution: Changchun University of Technology
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,090
OpenAlex ID: W4403203534