Conventional non-contact thread measurement methods face limitations in addressing profile occlusion distortion caused by the helix angle and in accurately quantifying critical geometric errors such as pitch diameter runout. This paper, from a metrological viewpoint, proposes a novel high-precision collaborative measurement methodology that integrates multimodal machine vision with an optical micrometer. The core innovation of this approach is twofold: first, it combines a traceable physical datum provided by the optical micrometer with highresolution profile analysis from machine vision, enabling the synergistic measurement of both macroscopic dimensions and microscopic features. Second, a mathematical model is developed and validated to compensate for the occlusion distortion inherent in vertical projection, thereby eliminating a significant systematic measurement error. The findings indicate a high degree of precision, with repeatability standard deviations of 1.3 mu m for the major diameter and 1.25 mu m for the pitch diameter. Furthermore, the system demonstrates exceptional efficiency, capable of completing a full-dimensional inspection of a single cross-section in under one second. These combined strengths in precision and speed confirm that the proposed methodology meets the stringent requirements for high-throughput industrial thread inspection.

Off-axis aspherical mirrors are widely used in optical systems and precision measuring instruments, whereas off-axis aspherical mirrors with large sizes and off-axis are used in large optical systems such as astronomical telescopes and radio telescopes. However, if the off-axis amount of an off-axis aspherical mirror exceeds the capability of the machine tool, traditional rotary-turning machining methods are not applicable, and advanced computerized numerical control (CNC) machining methods, such as the slow-tool-servo method, must be implemented. This article proposes a non-conventional offset (NCO) fabrication method based on slow-tool-servo single-point diamond turning for machining off-axis aspherical surfaces with large off-axis amounts. This method is theoretically applicable to the machining of off-axis aspherical surfaces with any off-axis amount. NCO fabrication is a simpler and more efficient path-planning solution for machining individual off-axis parabolic surfaces. In addition, corresponding solutions for other types of aspherical surfaces are proposed using the NCO method. The turning depths of workpieces with different off-axis amounts at the same machining position are analyzed and compared. A specific measurement scheme for the NCO method is presented, and the experimental results indicate that the PV and RMS form errors are 0.658 mu m and 60 nm, respectively. This work demonstrates that the NCO method can effectively deal with the machining challenges of off-axis aspherical structures with large off-axis amounts.

The rotary shaft is a critical component in aeroengines, serving to support transmission parts, transmit torque, and withstand loads. With the rapid advancement of the aviation industry, the precision requirements for shaft components have become increasingly stringent. Traditional measurement techniques often fall short in achieving high-precision and high-efficiency automated measurements. To address this challenge, this study develops a vertical digital measurement system for shafts. Leveraging advanced technologies such as machine vision, optical detection, and deep learning, the system employs a multi-sensor collaborative approach to investigate hardware design, error compensation, the implementation and application of various measurement techniques, and the development of measurement software. By analyzing hundreds of macro and micro characteristics of complex rotary shafts, the system achieves automated measurement with single clamping. The system features an axial measurement range of 1300 mm, a radial measurement range of 120 mm, a measurement cycle of less than 15 min, and achieves an expanded uncertainty of 1.2 mu m for macro-contour profiling and +/- 1 mu m for runout measurement.

Accurate 3D metrology of aero-engine cooling holes is critical for performance evaluation in advanced manufacturing, yet conventional methods often fail to meet high-throughput requirements. We propose FS-SFF, a novel deep learning framework for automated, high-precision 3D metrology based on the Shape-from-Focus principle. The framework integrates a custom optical system with a new reconstruction engine, FS-Net, featuring specialized modules to address the limitations of classical techniques, particularly sensitivity to noise and ambiguity on low-texture surfaces. Furthermore, an integrated analysis module incorporating deterministic aerospace engineering standards automatically quantifies critical engineering parameters from the reconstructed point clouds. Experiments demonstrate that FS-Net achieves a mean Intersection over Union (mIoU) of 88.88% on the ACHIS dataset, enabling geometric measurements with a mean absolute diameter error of 2.87 & micro;m. This corresponds to a 42% reduction in measurement error compared to conventional approaches. Achieving high-precision reconstructions at a throughput of 40 frames per second, FS-Net enables automated, standards-compliant quality control, facilitating its reliable deployment in the aerospace manufacturing industry. The code and dataset are publicly available at https://github.com/leilixjtu19-ctrl/FS-SFF.

The metrological accuracy of five-axis optical microscopy systems is critically constrained by coupled optomechanical errors, which induce both mechanical positioning deviations and degradation of optical imaging fidelity. This study introduces a systematic framework to model and compensate for these coupled errors. We advance the classic multi-body kinematic theory by developing an integrated error model that incorporates optical performance metrics, specifically defocus, directly within the kinematic chain. The model, formulated using homogeneous coordinate transformations and characterized by a detailed Jacobian sensitivity matrix, was populated with data from high-precision laser interferometers. Experimental validation on a calibrated artifact confirmed the framework's effectiveness. The compensation strategy reduced the aggregate standard deviation of form errors from 6.5 mu m to a mere 0.17 mu m, a 97.4 % improvement in measurement consistency. This dramatic enhancement validates the framework's capacity to elevate the system to a high-fidelity metrological instrument.

The topographical characteristics of diamond grinding wheels are critical to their performance, yet their quantification presents significant metrological challenges. Focus Variation (FV) microscopy is a promising technique for this purpose, but standard focus evaluation algorithms often fail when encountering the optically complex surfaces of grinding wheels, which feature specular reflections from abrasive grains and low-contrast textures from bonding materials. This leads to inaccurate 3D reconstructions and unreliable parameter extraction. To address this limitation, this paper proposes an enhanced FV methodology centered on a novel MultiGradient Adaptive Focus Evaluator (MGAF). The MGAF is specifically designed to improve the robustness and accuracy of focal plane detection on such challenging surfaces. Furthermore, a post-processing pipeline integrating a Multi-Scale Clustered Profiling technique based on Dual-Tree Complex Wavelet Transform is employed for the precise quantification of surface roughness and abrasive grain distribution. Comprehensive validation was performed against commercial system. The results demonstrate excellent agreement, with Sa discrepancies below 10% and a relative error in abrasive grain density quantification of less than 5%. This work validates the proposed methodology as a robust and accurate tool for the in-depth characterization of diamond grinding wheels, paving the way for improved quality control and process optimization.

88 % the stringent operational demands of aero-engine rotary shafts present a significant metrological conflict: traditional contact methods like CMM are prohibitively slow, while single-modality optical techniques struggle to reconcile the trade-off between large measurement volumes and high resolution. This paper introduces a novel hybrid optical metrology framework that fundamentally resolves this conflict through a synergistic fusion of complementary non-contact modalities and advanced algorithms. For meter-scale axial metrology, a telecentric imaging module is coupled with a robust affine transformation-based stitching algorithm that maintains metrological integrity for distortion-free axial and hole measurements across a 1300 mm range. A key methodological innovation is our mathematical datum referencing algorithm, which computationally establishes a virtual rotation axis directly from workpiece features. This approach delivers high-accuracy runout measurements that are robust against mounting inaccuracies, a critical source of error in conventional systems. Concurrently, an optical micrometer subsystem provides high-speed, sub-micron quantification of radial dimensions and dynamic geometric tolerances. By fusing data from these complementary sensors within a single clamping operation, the system achieves synchronized evaluation of over 200 geometric parameters. Experimental validation against a metrological CMM demonstrates a high radial repeatability, with a standard deviation (sigma) of 0.8 mu m. The system achieves an expanded uncertainty of 1.2 mu m for macro-contour profiling and +/- 1 mu m for runout measurement. Critically, this synergistic architecture enables a full shaft inspection in 15 min, reducing the measurement cycle time by 88 % compared to a benchmark commercial system while achieving metrologically equivalent accuracy. This work therefore provides a high-efficiency, high-fidelity solution, validated by novel algorithms and a multi-sensor architecture, for the comprehensive geometric evaluation of largescale, complex shafts, paving the way for 100 % in-line quality control in advanced manufacturing.

Stereo deflectometry is a promising method for the in-situ measurement of complex freeform surfaces. Calibration of the system is fundamental for accurate reconstruction of surfaces. In this study, we present a stepwise optimization algorithm for system parameter calibration that requires only a single spherical mirror pose. The proposed method addresses the issue of matrix ill-conditioning typically caused by an excessive number of variables in the optimization process. By exploiting the characteristics of stereo deflectometry, the parameter space is decomposed and optimized in stages, enabling effective and stable calibration. Experimental results confirm that the method accurately calibrates all relevant parameters, providing an efficient and practical calibration approach.

In precision machining, the geometry of the milling cutter plays a critical role in determining the quality of the milled surface. Detecting cutter wear is essential for maintaining machining quality. However, traditional measurement methods, primarily relying on signals and images, are inadequate for direct assessment of the wear area owing to the cutter's complex structure. This study introduces a 3D reconstruction method based on a sequence of microscopic images. To improve the measurement of milling tool morphology, a multi-partition Laplace focus evaluation operator is proposed to determine the optimal focus position of the images. Furthermore, a Gaussian fitting method with hierarchical window multi-scale fusion is implemented to efficiently identify the optimal focus position. Compared to measurements obtained using a Sensofar microscope, the standard deviation of the reconstructed wear area morphology is less than 1.65 %, and that of the milling cutter diameter is less than 0.010 mm. These findings confirm the accuracy of the proposed method in evaluating milling cutter wear.

Industrial defect segmentation is critical for manufacturing quality control. Due to the scarcity of training defect samples, few-shot semantic segmentation (FSS) holds significant value in this field. However, existing studies mostly apply FSS to tackle defects on simple textures, without considering more diverse scenarios. This paper aims to address this gap by exploring FSS in broader industrial products with various defect types. To this end, we contribute a new real-world dataset and reorganize some existing datasets to build a more comprehensive few-shot defect segmentation (FDS) benchmark. On this benchmark, we thoroughly investigate metric learning-based FSS methods, including those based on meta-learning and those based on Vision Foundation Models (VFM). We observe that existing meta-learning-based methods are generally not well-suited for this task, while VFMs hold great potential. We further systematically study two types of VFMs, including upstream representation learning models and downstream SAM (Segment anything) series models. We propose a novel training-free FDS method, called FM-SAM, which is based on feature matching combined with FastSAM for refinement. It demonstrates convincing segmentation performance while maintaining high efficiency. In addition, we find that SAM2 can directly perform effective FDS end-to-end through its video tracking mode. The contributed dataset and code are available at: https://github.com/liutongkun/GFDS.

The complex surface characteristics and misalignment of the geometric and optical axes of off-axis aspherical surfaces pose major challenges in the work to develop a measurement technique that is accurate, effective, and simple. Phase measurement deflectometry (PMD) is a more convenient and cost-effective method than interferometry for measuring specular objects. Nevertheless, monocular PMD necessitates complicated processes or costly additional equipment to resolve height-slope problems, whereas binocular PMD requires more time-consuming matching calculations. In this article, a novel iterative strategy based on monocular PMD technology with an automatically updated seed point during the iterative process is proposed, which only requires an approximate estimation about the height of one point (either the lowest, middle, or highest) on the measured off-axis aspheric component. The lowest, middle, or highest point of the reconstructed surface is used as the target of the auto-updated seed point in the iterative process, and the surface's overall height information is then updated until the difference between two neighboring surface profiles is less than a certain threshold. Finally, the off-axis aspherical surface result is sent out. The proposed monocular PMD method is more straightforward, cost-effective, and simple, requiring only basic and low-cost tools to estimate the initial height of the measured component without the need to obtain the seed point's 3-D coordinate. An experiment was performed to evaluate the feasibility and accuracy of the proposed method with the results of high-precision contact measurements as benchmarks. The results also indicated that even the uncertainty in the estimated initial height using common Vernier calipers has little effect on the measured surface shape, since it only leads to a small offset in the whole surface location.

With the continuous improvement of remote sensing satellite resolution, laser communication technology has gained significant traction. The pointing accuracy of ground-based laser communication terminals is critical for the stability of satellite–ground laser transmission links. To enhance the pointing accuracy of ground-based laser communication terminals, this study proposes a high-precision calibration methodology utilizing an error correction mathematical model. This approach complements traditional methods. The pointing errors of an alt-azimuth dual-axis laser communication terminal system are analyzed, and the principles and implementation processes of the error correction mathematical model are presented. Calibration experiments were conducted using an existing laser communication terminal test platform. Observation error data were obtained by comparing stellar observations with theoretical stellar positions, and error model parameters were fitted. Verification through stellar observations after model establishment and error correction showed that the mean open-loop pointing error can be controlled to approximately 5″ or less. Compared to traditional methods, accuracy can be improved by over 85%, demonstrating significant and highly accurate error correction effects and validating the proposed method.

The milling tool plays a pivotal role in the fabrication of components within the aerospace and various other industrial sectors. Consequently, it is essential to perform swift, accurate, and comprehensive evaluations of tool wear throughout the manufacturing and processing stages. Traditional methods for assessing tool wear often suffer from limitations due to their dependence on singular evaluation criteria and a lack of detailed wear information. To overcome these challenges, this study introduces a multi-dimensional tool wear detection system utilizing optical microscopy vision to capture fully-focused images of the tool and reconstruct the shape of the wear region, which is designed to enable efficient, high-precision, and holistic evaluation of wear parameters. Furthermore, Non-Subsampled Shearlet Transform (NSST) and an enhanced pulse coupled neural network (PCNN)are used to extract 3D depth information, which facilitate the creation of a high-precision tool depth map by mapping high-frequency subbands to different depth levels while simultaneously obtaining the fully-focused image. Additionally, an inspection criterion is established that encompasses a multi-dimensional evaluation of wear metrics, including wear value, area, and volume. Compared to standard equipment, the error of wear value was found to be less than 0.005 mm, and the error rate of area, and volume was less than 2.5 %. Experimental results demonstrate that the proposed method offers more comprehensive assessment metrics for evaluating tool wear. It can be used to offer valuable feedback of tool state for machining processes.

The evaluation of the film cooling performance of turbine blades necessitates the analysis of the geometric parameters of the cooling holes. Nevertheless, obtaining precise measurements poses a challenge due to the intricate structure of the cooling holes, especially the fan-shaped cooling holes with tiny diameters, to which conventional measurement methods cannot be applied. The paper presents a novel method for measuring the multi-view internal fusion morphology of cooling holes, aiming to evaluate geometric parameters accurately, which utilizes the optimal beluga whale optimization (OBWO) algorithm and a weighted variance algorithm that relies on the depth information of point clouds. Experiments simulating the cooling hole process are conducted to validate the proposed methodology. The findings indicate that the method effectively recovers the internal 3D shape of the cooling hole. The above method is also employed to quantify intricate fan-shaped cooling apertures on physical turbine blades. The cooling hole diameter exhibits a maximum error of less than 9.5 mu m, while the hole axis angle demonstrates a maximum deviation of less than 0.87. The proposed method accomplishes the measurement of the internal morphology of the cooling hole and evaluates its geometric parameters.

Microscopic images of surfaces can be used for non-contact roughness measurement by visual methods. However, the images are usually acquired manually and need to be as sharp as possible, which limits the general application of the method. This manuscript provides an automatic roughness measurement method that can apply to automatic industrial sites. This method first automatically acquires the sharpest image and then feeds the image into a convolutional neural network (CNN) model for roughness measurement. In this method, the weighted window enhanced sharpness evaluation algorithm based on the sharpness evaluation function is proposed to automatically extract the sharpest image. Then, a CNN model, CFEN, suitable for the roughness measurement task was designed and pre-trained. The results demonstrate that the measurement accuracy of the method reaches 91.25% and the time is within a few seconds. It is proved that the method has high accuracy and efficiency and is feasible in practical applications.

In order to measure the inner diameter of the film cooling hole in turbine blades, the 3D point cloud reconstruction method was studied in this paper. A five-axis motion platform and microscopic vision probe were used to obtain the focusing image sequence of the film cooling hole. Image processing techniques such as image enhancement and Image noise reduction are used to improve the quality of the focusing image sequence. Then, edge detection and threshold selection are used to extract clear pixel points representing the inner wall information of the film cooling hole in the focusing image sequence. Based on the grating ruler readings of the five axis motion platform during image sequence collection, different depth information of the measured film cooling hole is transformed. Combined with the clear pixel points, a three-dimensional point cloud reflecting the internal morphology information of the film cooling hole is reconstructed. Finally, the three-dimensional point cloud is processed to obtain parameter information such as the inner diameter of the measured film cooling hole.

The parameters of cooling holes on aerospace engine blades, including diameter, angle, and depth, critically influence the blades' cooling efficiency and overall performance. However, due to the intricate structure of the cooling holes, conventional measurement methods often fall short in terms of accuracy and efficiency. Addressing these challenges, this paper employs Shape from Focus (SFF) technology to achieve highprecision measurement of cooling hole diameters. This approach entails capturing a series of images of the cooling holes using an advanced image acquisition platform. To enhance measurement accuracy and efficiency, the Laplace operator is utilized to extract clear regions from the image sequence. This process is further refined by incorporating Region of Interest (ROI) algorithms to select pertinent data related to the cooling holes, thereby facilitating the 3D reconstruction of these features. Compared to traditional methods, this technique significantly improves computational efficiency and processing speed while enhancing the accuracy and stability of image processing. The method achieves a measurement error of less than 0.005 mm and a relative error of less than 1.5% when measuring the diameters of cooling holes, thereby providing robust support for industrial inspection and quality control. Moreover, it holds broad application prospects and substantial practical value.