Zijian Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zijian Zhu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Chenyang Zhao
School of Robotics and Advanced Manufacture, Harbin Institute of Technology, Shenzhen, China
14 shared publications - Weien Yu
School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China
4 shared publications - Yueping Xi
School of Robotics and Advanced Manufacture, Harbin Institute of Technology at Shenzhen, Shenzhen, China
4 shared publications - Qinghong Wan
Harbin Institute of Technology (Shenzhen)
3 shared publications - Yang Yang
School of Robotics and Advanced Manufacture Harbin Institute of Technology (Shenzhen) Shenzhen China
3 shared publications - Honglu Li
Harbin Institute of Technology School of Mechanical Engineering and Automation Shenzhen 518055
2 shared publications
Latest Publications
- 30 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-14
A Defocus-Robust Encoded Pattern for Simultaneous Absolute Positioning and Parallelism Evaluation in High-Magnification Microscopy
Microscopy-based positioning techniques are applicable to precision manufacturing scenarios, such as semiconductor fabrication. High-magnification microscopic metrology (e.g., using a 100× objective) provides an extremely small pixel equivalent size, thereby offering a solid physical basis for achieving nanometer-level absolute positioning. However, the extremely shallow depth of field—typically a few hundred nanometers—makes spatially varying defocus, mainly induced by target–stage parallelism errors, a critical obstacle to robust large-range positioning. Existing encoded-pattern absolute positioning methods are therefore often demonstrated at low magnification, where defocus is less severe. This paper proposes an encoded-pattern method that enables both defocus-robust absolute positioning and in-situ installation parallelism evaluation under high magnification. For positioning, a sub-pixel edge-gradient scheme is used to estimate feature locations with reduced sensitivity to blur. For parallelism evaluation, the defocus-dependent blur magnitude is leveraged to quantify the target tilt. Experiments demonstrate a positioning resolution characterized by a standard deviation of 0.8 nm and an RMSE of 115.5 nm over a 3 mm × 3 mm range. Moreover, the proposed in-situ parallelism evaluation estimates the target tilt with a mean absolute error of 94.8 nm.
- 30 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-14
A Defocus-Robust Encoded Pattern for Simultaneous Absolute Positioning and Parallelism Evaluation in High-Magnification Microscopy
Microscopy-based positioning techniques are applicable to precision manufacturing scenarios, such as semiconductor fabrication. High-magnification microscopic metrology (e.g., using a 100× objective) provides an extremely small pixel equivalent size, thereby offering a solid physical basis for achieving nanometer-level absolute positioning. However, the extremely shallow depth of field—typically a few hundred nanometers—makes spatially varying defocus, mainly induced by target–stage parallelism errors, a critical obstacle to robust large-range positioning. Existing encoded-pattern absolute positioning methods are therefore often demonstrated at low magnification, where defocus is less severe. This paper proposes an encoded-pattern method that enables both defocus-robust absolute positioning and in-situ installation parallelism evaluation under high magnification. For positioning, a sub-pixel edge-gradient scheme is used to estimate feature locations with reduced sensitivity to blur. For parallelism evaluation, the defocus-dependent blur magnitude is leveraged to quantify the target tilt. Experiments demonstrate a positioning resolution characterized by a standard deviation of 0.8 nm and an RMSE of 115.5 nm over a 3 mm × 3 mm range. Moreover, the proposed in-situ parallelism evaluation estimates the target tilt with a mean absolute error of 94.8 nm.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Phase Error Suppression in Periodic Patterns: A High Precision Method for Industrial Periodic Array Feature Extraction and Calibration
In industrial micro-vision systems, challenges like mounting parallelism, mechanical vibration, and thermal drift often cause frequent and non-uniform defocus, which degrades traditional corner detection and complicates camera calibration. While phase-based techniques are robust to defocus, their accuracy is limited by phase extraction errors. To overcome these issues, this paper combines Phase Error Suppression (PES) with full-field phase modeling for robust feature localization and accurate camera calibration of periodic array patterns in high-magnification scenarios. PES method enhances phase extraction and localization accuracy by optimizing parameters like windowing and filtering bandwidth, and full-field phase modeling enables precise radial distortion estimation from a single image. Experiments validate our method's accuracy and robustness under high defocus, demonstrating average errors of 120.59 nm for 90 μm displacement, 190 nm for periodic pattern dimension measurement. Its effective application in wafer solder joint pitch measurement yields an error of 491 nm, offering a reliable solution for high-precision micro-vision measurement and calibration.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Phase Error Suppression in Periodic Patterns: A High Precision Method for Industrial Periodic Array Feature Extraction and Calibration
In industrial micro-vision systems, challenges like mounting parallelism, mechanical vibration, and thermal drift often cause frequent and non-uniform defocus, which degrades traditional corner detection and complicates camera calibration. While phase-based techniques are robust to defocus, their accuracy is limited by phase extraction errors. To overcome these issues, this paper combines Phase Error Suppression (PES) with full-field phase modeling for robust feature localization and accurate camera calibration of periodic array patterns in high-magnification scenarios. PES method enhances phase extraction and localization accuracy by optimizing parameters like windowing and filtering bandwidth, and full-field phase modeling enables precise radial distortion estimation from a single image. Experiments validate our method's accuracy and robustness under high defocus, demonstrating average errors of 120.59 nm for 90 μm displacement, 190 nm for periodic pattern dimension measurement. Its effective application in wafer solder joint pitch measurement yields an error of 491 nm, offering a reliable solution for high-precision micro-vision measurement and calibration.
- 20 November 2025
- Published by Elsevier in Optics & Laser Technology
- Vol. 193
Dark-field feature position: A nanoscale control point extraction method for micro-vision calibration and measurement
- 20 November 2025
- Published by Elsevier in Optics & Laser Technology
- Vol. 193
Dark-field feature position: A nanoscale control point extraction method for micro-vision calibration and measurement
- 6 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
An Investigation of the Planar 3-DOF Nanoscale Positioning Measurement Method Based on Cross-Encoded Pattern
Precision positioning measurement technology based on micro-vision has important applications in high-end manufacturing fields such as semiconductors and machine tools. Currently, measurement methods based on pseudo-periodic patterns can break the limitations of periodic patterns and achieve a larger measurement range. However, existing methods have not considered the impact of defocus caused by large-range displacement or assembly misalignment on measurement accuracy. Additionally, the physical period of pseudo-periodic encoding units and the bit number are interdependent, which requires careful trade-offs when expanding the measurement range. To address these issues, this paper innovatively proposes a plane 3-DOF nanoscale positioning measurement method based on Adaptive Cross-Encoded Pattern (ACEP). By utilizing high-precision etching technology to ensure the center information of the patterns and encoding the size information of the pattern units, the proposed method mitigates the impact of defocus on measurement. A key advantage of the cross-encoded pattern design is its range scalability, which allows the pattern size to be adjusted to extend the measurement range. Experimental results show that the proposed method achieves a linear resolution of 20 nm and an angular resolution of 20 seconds, and the measurement accuracy remains stable over a 9 mm displacement range, demonstrating the method’s effectiveness and stability at nanoscale precision.
- 6 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
An Investigation of the Planar 3-DOF Nanoscale Positioning Measurement Method Based on Cross-Encoded Pattern
Precision positioning measurement technology based on micro-vision has important applications in high-end manufacturing fields such as semiconductors and machine tools. Currently, measurement methods based on pseudo-periodic patterns can break the limitations of periodic patterns and achieve a larger measurement range. However, existing methods have not considered the impact of defocus caused by large-range displacement or assembly misalignment on measurement accuracy. Additionally, the physical period of pseudo-periodic encoding units and the bit number are interdependent, which requires careful trade-offs when expanding the measurement range. To address these issues, this paper innovatively proposes a plane 3-DOF nanoscale positioning measurement method based on Adaptive Cross-Encoded Pattern (ACEP). By utilizing high-precision etching technology to ensure the center information of the patterns and encoding the size information of the pattern units, the proposed method mitigates the impact of defocus on measurement. A key advantage of the cross-encoded pattern design is its range scalability, which allows the pattern size to be adjusted to extend the measurement range. Experimental results show that the proposed method achieves a linear resolution of 20 nm and an angular resolution of 20 seconds, and the measurement accuracy remains stable over a 9 mm displacement range, demonstrating the method’s effectiveness and stability at nanoscale precision.
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
Microscale 3-D Rotation Estimation Based on Deflected Defocused Feature Extraction
In micro-electromechanical systems, measuring three-dimensional rotation presents challenges such as an unclear rotation center and poor orthogonality. Micro-vision-based rotation estimation is limited by a shallow imaging depth-of-field, leading to narrow measurement range and low precision. This paper introduces a novel method called deflected defocused feature extraction (DFE) to address these issues and enable accurate microscale rotation estimation. DFE extends the measurement range from 0.03 rad to 0.92 rad while achieving a precision of 0.1179 mrad for out-of-plane angles. The edge defocus model incorporating deflection angles is derived based on microscopic imaging principles. A general circular array is fabricated and a robust defocus control point extraction method is proposed. Finally, three-dimensional angular displacement measurement is accomplished via homography matrix decomposition. Experimental results demonstrate a resolution of 0.17 mrad, 0.17 mrad, and 0.0087 mrad for the three axes, respectively. The range-resolution ratios are found to be 104, 104, and 106, respectively.
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
Microscale 3-D Rotation Estimation Based on Deflected Defocused Feature Extraction
In micro-electromechanical systems, measuring three-dimensional rotation presents challenges such as an unclear rotation center and poor orthogonality. Micro-vision-based rotation estimation is limited by a shallow imaging depth-of-field, leading to narrow measurement range and low precision. This paper introduces a novel method called deflected defocused feature extraction (DFE) to address these issues and enable accurate microscale rotation estimation. DFE extends the measurement range from 0.03 rad to 0.92 rad while achieving a precision of 0.1179 mrad for out-of-plane angles. The edge defocus model incorporating deflection angles is derived based on microscopic imaging principles. A general circular array is fabricated and a robust defocus control point extraction method is proposed. Finally, three-dimensional angular displacement measurement is accomplished via homography matrix decomposition. Experimental results demonstrate a resolution of 0.17 mrad, 0.17 mrad, and 0.0087 mrad for the three axes, respectively. The range-resolution ratios are found to be 104, 104, and 106, respectively.
- 13 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Influence Mechanism of the Surface Topography on the Positioning Measurement Accuracy of Micro-Vision Pattern Targets
In micro-vision-based positioning, designing special patterns as positioning targets is key to improving the positioning accuracy and measurement range. Many scholars, however, improve the accuracy by processing different targets and designing different algorithms while ignoring the influence of microtopography/nanotopography distribution on positioning measurement. From this perspective, this article focuses on two micro-surface processing methods, ultraprecision machining, and photolithography, to explore the mechanism of the influence of the surface topography on the positioning measurement accuracy. The influence of the processing technology on the positioning accuracy and the optimal processing parameters are obtained via exploring the influence of the processing technology on the 3-D topography, the influence of the 3-D topography on the imaging grayscale, and the influence of the grayscale distribution on positioning measurement. The experimental results are consistent with the theoretical analysis. For ultraprecision machining, there is an optimal value for the cutting speed and the positioning accuracy is proportional to the cutting depth. For photolithography, the positioning accuracy is proportional to the surface reflectivity difference. Under the optimal parameter configuration, the standard deviation (STD) of repeated positioning can reach 1.69 nm.
- 13 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Influence Mechanism of the Surface Topography on the Positioning Measurement Accuracy of Micro-Vision Pattern Targets
In micro-vision-based positioning, designing special patterns as positioning targets is key to improving the positioning accuracy and measurement range. Many scholars, however, improve the accuracy by processing different targets and designing different algorithms while ignoring the influence of microtopography/nanotopography distribution on positioning measurement. From this perspective, this article focuses on two micro-surface processing methods, ultraprecision machining, and photolithography, to explore the mechanism of the influence of the surface topography on the positioning measurement accuracy. The influence of the processing technology on the positioning accuracy and the optimal processing parameters are obtained via exploring the influence of the processing technology on the 3-D topography, the influence of the 3-D topography on the imaging grayscale, and the influence of the grayscale distribution on positioning measurement. The experimental results are consistent with the theoretical analysis. For ultraprecision machining, there is an optimal value for the cutting speed and the positioning accuracy is proportional to the cutting depth. For photolithography, the positioning accuracy is proportional to the surface reflectivity difference. Under the optimal parameter configuration, the standard deviation (STD) of repeated positioning can reach 1.69 nm.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (4) , 6783-6793
Micro-Vision-Based 3-DOF Full-Stroke Nanoscale Positioning Method for the Wafer Stage
The quality of semiconductor manufacturing and processing technology is determined by the precision and efficiency of the wafer stage positioning measurement. However, traditional measurement sensors such as laser interferometers and grating rulers are expensive to calibrate and have poor orthogonality. To address this, this paper presents a micro-vision-based wafer stage positioning system that employs a line-feature-based coarse-to-fine (LFCF) method for global planar three-degrees-of-freedom positioning via the wafer pattern. Line features are extracted from the wafer to estimate the angular displacement. An image database is established, and the residual neural network is improved for coarse positioning to eliminate accumulated errors. The fast and robust feature-based positioning (FRFP) algorithm is proposed for feature extraction and matching, and the fine position is obtained through the constraint equation of feature point pairs. This paper also builds a micro-vision-based wafer positioning system to verify the precision and efficiency of the approach. The results demonstrate that the system achieves a precision of 131.4 nm, 82.2 nm, and 0.0015 rad in the x, y, and θ directions, respectively, at an image processing frame rate of 5 Hz.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (4) , 6783-6793
Micro-Vision-Based 3-DOF Full-Stroke Nanoscale Positioning Method for the Wafer Stage
The quality of semiconductor manufacturing and processing technology is determined by the precision and efficiency of the wafer stage positioning measurement. However, traditional measurement sensors such as laser interferometers and grating rulers are expensive to calibrate and have poor orthogonality. To address this, this paper presents a micro-vision-based wafer stage positioning system that employs a line-feature-based coarse-to-fine (LFCF) method for global planar three-degrees-of-freedom positioning via the wafer pattern. Line features are extracted from the wafer to estimate the angular displacement. An image database is established, and the residual neural network is improved for coarse positioning to eliminate accumulated errors. The fast and robust feature-based positioning (FRFP) algorithm is proposed for feature extraction and matching, and the fine position is obtained through the constraint equation of feature point pairs. This paper also builds a micro-vision-based wafer positioning system to verify the precision and efficiency of the approach. The results demonstrate that the system achieves a precision of 131.4 nm, 82.2 nm, and 0.0015 rad in the x, y, and θ directions, respectively, at an image processing frame rate of 5 Hz.
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
Galaxy Microstructure: An Absolute Encoding Method for Planar 3-DOF Nanoscale Positioning Measurement
Micro-vision-based precision positioning measurement technology has emerged as a pivotal technology in the manufacturing of high-end equipment, including semiconductors and machine tools. Among the critical parameters, scholars consistently focus on the measurement range and resolution. However, these factors are often interdependent due to the constraints imposed by target pattern preparation accuracy, design methodology, and limitations of image registration algorithms. This paper addresses these challenges by introducing a novel encoding pattern with ultra-precision diamond cutting, named the galaxy microstructure. This innovation achieves plane three-degrees-of-freedom positioning measurement, surpassing the range-to-resolution ratio of 10 7 . The clear imaging mechanism and distinctive coding method, grounded in ultra-precision processing, ensure the accuracy in pattern preparation and extend the measurement range. The robustness and accuracy of the coding unit feature positioning method guarantee high resolution and efficient positioning. Experimental validation confirms that the galaxy microstructure positioning method attains an impressive range-to-resolution ratio of 3.6×10 7 and boasts a processing efficiency of 20 Hz.
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
Galaxy Microstructure: An Absolute Encoding Method for Planar 3-DOF Nanoscale Positioning Measurement
Micro-vision-based precision positioning measurement technology has emerged as a pivotal technology in the manufacturing of high-end equipment, including semiconductors and machine tools. Among the critical parameters, scholars consistently focus on the measurement range and resolution. However, these factors are often interdependent due to the constraints imposed by target pattern preparation accuracy, design methodology, and limitations of image registration algorithms. This paper addresses these challenges by introducing a novel encoding pattern with ultra-precision diamond cutting, named the galaxy microstructure. This innovation achieves plane three-degrees-of-freedom positioning measurement, surpassing the range-to-resolution ratio of 10 7 . The clear imaging mechanism and distinctive coding method, grounded in ultra-precision processing, ensure the accuracy in pattern preparation and extend the measurement range. The robustness and accuracy of the coding unit feature positioning method guarantee high resolution and efficient positioning. Experimental validation confirms that the galaxy microstructure positioning method attains an impressive range-to-resolution ratio of 3.6×10 7 and boasts a processing efficiency of 20 Hz.
- 4 October 2024
- Published by ASME International in Journal of Manufacturing Science and Engineering
- Vol. 147 (2) , 1-41
Modeling and Experimental Investigation of Surface Generation in Diamond Micro-Chiseling
The investigation of diamond micro-chiseling (DMC) has been studied as a machining process for generating specific surface microstructures. The current research mainly focuses on the processing performance of DMC in different applications without exploring the DMC surface generation process from the principle. This article studies the DMC surface generation process under different processing parameters through both theoretical and experimental parts. In the theoretical part, an accurate surface generation model is created that considers the effects of geometrical relationships, material removal mechanisms, minimum uncut chip thickness, and dynamics. The geometric model of the diamond chiseling tool is described, the trajectory between the tool and workpiece is built, and the minimum uncut chip is considered and established. Regarding dynamic factors, a spring-mass damping vibration system of DMC is established, and a finite element model is built to analyze the cutting force. In the experimental part, DMC surface generation experiments are conducted to explore the material removal mechanism and compare surface topography. Results show that our simulation model can estimate DMC topography with an error of less than 0.5 µm. Simultaneously, the optimal DMC strategy is obtained through experiments.
- 4 October 2024
- Published by ASME International in Journal of Manufacturing Science and Engineering
- Vol. 147 (2) , 1-41
Modeling and Experimental Investigation of Surface Generation in Diamond Micro-Chiseling
The investigation of diamond micro-chiseling (DMC) has been studied as a machining process for generating specific surface microstructures. The current research mainly focuses on the processing performance of DMC in different applications without exploring the DMC surface generation process from the principle. This article studies the DMC surface generation process under different processing parameters through both theoretical and experimental parts. In the theoretical part, an accurate surface generation model is created that considers the effects of geometrical relationships, material removal mechanisms, minimum uncut chip thickness, and dynamics. The geometric model of the diamond chiseling tool is described, the trajectory between the tool and workpiece is built, and the minimum uncut chip is considered and established. Regarding dynamic factors, a spring-mass damping vibration system of DMC is established, and a finite element model is built to analyze the cutting force. In the experimental part, DMC surface generation experiments are conducted to explore the material removal mechanism and compare surface topography. Results show that our simulation model can estimate DMC topography with an error of less than 0.5 µm. Simultaneously, the optimal DMC strategy is obtained through experiments.