Combined active triangulation, morphology scheme for active shape retrieval

Abstract

Purpose

The purpose of this paper is to present a combined scheme of active laser‐based triangulation and a morphological edge detection, to quantify features dimensions (width and depth) over smooth plastic surfaces. The inspected target is an automotive polyethylene, optically black fender, with average roughness of 1μm, while the sought features are mold sinks with depth variations of 0.3‐1.0 mm and a width of ∼3 mm.

Design/methodology/approach

Several non‐contact scanning and reconstruction optical modalities are discussed along with its associated noise for current application; such techniques include interferometry and triangulation. The proposed system projects a linear coherent illumination and scans its reflection profile to infer the reflected component position using trigonometry and its width using a Sobel morphological operator. The retrieved dimensions are then reconstructed into feature boundaries that enable a depth and width quantification.

Findings

The proposed setup and processing are validated through experimental scans of actual molded fenders with artificial deviations. The proposed system accuracy is then analyzed and its spatial resolution is found to be 0.14 mm using the current charge‐coupled device format of 640 × 480. Additionally the proposed system is benchmarked against a commercial 3D stereo‐based scanner; proposed system proved to be more accurate with faster scanning rates.

Research limitations/implications

The post processor combines the predicted width and depth values in real‐time to synthesize a 3D surface profile for sought features, with a resolution of 0.14 mm, an accuracy of 0.09 mm, and a repeatability of 0.11 mm. The proposed scheme is customized to current laser illuminant and to the plastic surface profile, which further modifications for other illuminations and roughness values.

Originality/value

New image processing code is used to reduce the laser speckle effect and the moving mechanism vibration using a moving Gaussian illuminant and a double thresholding scheme coupled with sequential averaging, respectively.

Citation

Omar, M., Zhou, Y., Planting, E., Parvataneni, R. and Hung, S. (2009), "Combined active triangulation, morphology scheme for active shape retrieval", Sensor Review, Vol. 29 No. 3, pp. 233-239. https://doi.org/10.1108/02602280910967648