Multi-Scale Adaptive Edge Detection Measures Ruby Balls With 1.6 Micrometer Error

A high-precision, non-contact optical measurement method has demonstrated stable dimensional inspection of precision spherical components by coupling annular dark-field imaging with multi-scale adaptive edge detection. In experimental evaluations using a 5 mm ruby ball sample, the integrated system attained a mean absolute error of 1.6 μm alongside a standard deviation of 0.7 μm, establishing a robust protocol for inspecting high-tolerance industrial sphere targets without surface contact.

The software framework subsequently applies multi-scale gradient-response fusion combined with entropy-guided local thresholding. This combination improves outer-contour extraction under conditions of non-uniform contrast and noise.

To achieve dimensional accuracy beyond standard sensor pixel grids, the technique incorporates cubic-spline interpolation for sub-pixel edge localization along the extracted contour. The filtered boundary coordinates are then resolved into final spherical diameters using geometric-distance circle fitting, minimizing the influence of outlier data points and generating dependable circular boundary fits across the inspected samples.

Under controlled noise conditions with a signal-to-noise ratio of 20 dB, the proposed measurement method maintained a dimensional error of 2.4 μm. This performance demonstrated significant precision gains over alternative computational methods tested under identical noise levels. Specifically, the evaluation revealed that the Hessian-matrix edge detection approach yielded a measurement error of 6.5 μm, while a conventional pixel-level regionprops baseline produced an error of 11.8 μm under the same 20 dB noise environment.

Beyond tests on single ruby spheres, experimental verification was expanded across multiple engineering materials and component sizes to confirm operational stability. Sun evaluated spherical specimens ranging from 3 mm to 12 mm in diameter, incorporating ruby, ceramic, and cemented-carbide materials. The consistent performance observed across varying component dimensions, material reflectivities, and imaging conditions indicates that combining dark-field illumination with adaptive multi-scale edge filtering provides an effective, non-destructive approach for the quality verification and dimensional inspection of precision spherical parts.

Subject of Research: Optical non-contact dimensional measurement of precision spherical components using multi-scale adaptive edge detection and annular dark-field imaging

Article Title: A high-precision measurement method for ruby balls based on multi-scale adaptive edge detection

Article References: Sun, W. (2026). A high-precision measurement method for ruby balls based on multi-scale adaptive edge detection. Discover Applied Sciences. https://doi.org/10.1007/s42452-026-09866-6

Image Credits: AI Generated

DOI: 10.1007/s42452-026-09866-6

Keywords: dimensional inspection, edge detection, annular dark-field imaging, ruby balls, sub-pixel localization, cubic-spline interpolation, RANSAC screening, circle fitting, geometric distance, optical measurement, non-contact metrology, precision spheres, ceramic balls, cemented-carbide balls, entropy thresholding, gradient response fusion