This paper shows the application of several learning-based image classification techniques to conformity check, which is a common problem in industrial visual inspection. The approaches are based on processing 2D images. First, a classification pipeline has been developed. An effort has been invested into choosing an appropriate classifier. First experiment was performed with HoG features (Histogram Of Gradient) and Support Vector Machine (SVM). Further, to improve accuracy, we employed a bag of visual words (BoVW) and ORB detector for extracting features that we further use to build our dictionary of visual words. The final solution uses features extracted by passing an image through a pre-trained deep convolutional neural network Inception. Using these features a SVM classifier was trained and high accuracy was obtained. To augment our image data set, different transformations such as zoom and shearing were applied. Promising results were obtained which shows that state-of-the-art deep learning classification techniques can be successfully employed in the visual industrial inspection field.

This paper describes the complete dimensioning of a method and apparatus dedicated to the automatic detection of defects on optical mirror-like components called optical solar reflector (OSR), which are used to ensure the thermal control of satellites. First, the requirements of the method are defined according to the properties of the OSR (multilayer and semitransparent) and the defects to be detected. The defect dimensions are then measured and exhibit very dispersed micrometric values. Second, the optical characterization of the OSR and its sublayers is conducted, enabling the dimensioning of the machine with regard to the localization of the defect (surface, interface, or volume). Third, a semi-industrial prototype is developed and evaluated through two kinds of experiments. The first experiment illustrates the production capacity of the prototype and the second the measurement capability. Results showed that the prototype exceeded the initial requirements and could be used to replace the currently used control with confidence.

In this paper, we propose a new approach to detect and inspect airplane tires. We use normalized correlation based template matching to detect tires and GrabCut segmentation method to extract them from the background. We inspect the tires condition by investigating the tire treads. Ridge based intensity profiling gives satisfying inspection results.

Abstract. This paper deals with an automated preflight aircraft inspection using a pan-tilt-zoom camera mounted on a mobile robot moving autonomously around the aircraft. The general topic is image processing framework for detection and exterior inspection of different types of items, such as closed or unlatched door, mechanical defect on the engine, the integrity of the empennage, or damage caused by impacts or cracks. The detection step allows to focus on the regions of interest and point the camera toward the item to be checked. It is based on the detection of regular shapes, such as rounded corner rectangles, circles, and ellipses. The inspection task relies on clues, such as uniformity of isolated image regions, convexity of segmented shapes, and periodicity of the image intensity signal. The approach is applied to the inspection of four items of Airbus A320: oxygen bay handle, air-inlet vent, static ports, and fan blades. The results are promising and demonstrate the feasibility of an automated exterior inspection.

This paper deals with the inspection of an airplane using a Pan-Tilt-Zoom camera mounted on a mobile robot moving around the airplane. We present image processing methods for detection and inspection of four different types of items on the airplane exterior. Our detection approach is focused on the regular shapes such as rounded corner rectangles and ellipses, while inspection relies on clues such as uniformity of isolated image regions, convexity of segmented shapes and periodicity of the image intensity signal. The initial results are promising and demonstrate the feasibility of the envisioned robotic system.

We focus on quality control of mechanical parts in aeronautical context using a single pan-tilt-zoom (PTZ) camera and a computer-aided design (CAD) model of the mechanical part. We use the CAD model to create a theoretical image of the element to be checked, which is further matched with the sensed image of the element to be inspected, using a graph theory-based approach. The matching is carried out in two stages. First, the two images are used to create two attributed graphs representing the primitives (ellipses and line segments) in the images. In the second stage, the graphs are matched using a similarity function built from the primitive parameters. The similarity scores of the matching are injected in the edges of a bipartite graph. A best-match-search procedure in the bipartite graph guarantees the uniqueness of the match solution. The method achieves promising performance in tests with synthetic data including missing elements, displaced elements, size changes, and combinations of these cases. The results open good prospects for using the method with realistic data. (C) 2015 SPIE and IS&T

Stereo digital image correlation (also called 3D DIC) is a common measurement technique in experimental mechanics for measuring 3D shapes or 3D displacement/strain fields, in research laboratories as well as in industry. Nevertheless, like most of the optical full-field measurement techniques, 3D DIC suffers from a lack of information about its metrological performances. For the 3D DIC technique to be fully accepted as a standard measurement technique it is of key importance to assess its measurement uncertainties.

ABSTRACTWe report on the main results of a collaborative work devoted to the study of the uncertainties associated with Digital image correlation techniques (DIC). More specifically, the dependence of displacement measurement uncertainties with both image characteristics and DIC parameters is emphasised. A previous work [Bornert et al. (2009) Assessment of digital image correlation measurement errors: methodology and results. Exp. Mech. 49, 353–370] dedicated to situations with spatially fluctuating displacement fields demonstrated the existence of an ‘ultimate error’ regime, insensitive to the mismatch between the shape function and the real displacement field. The present work is focused on this ultimate error. To ensure that there is no mismatch error, synthetic images of in‐plane rigid body translation have been analysed. Several DIC softwares developed by or in use in the French community have been used to explore the effects of a large number of settings. The discrepancies between DIC evaluated displacements and prescribed ones have been statistically analysed in terms of random errors and systematic bias, in correlation with the fractional part τ of the displacement component expressed in pixels. Main results are as follows: (i) bias amplitude is almost always insensitive to subset size, (ii) standard deviation of random error increases with noise level and decreases with subset size and (iii) DIC formulations can be split up into two main families regarding bias sensitivity to noise. For the first one, bias amplitude increases with noise while it remains nearly constant for the second one. In addition, for the first family, a strong dependence of random error with τ is observed for noisy images.

The wood plastic composites studied in this work are composed of high-density polyethylene matrix and Pinus pinaster wood fibers. Despite some interesting intrinsic properties, this wood plastic composite has limited mechanical properties because of the incompatibility between the polar hydrophilic fibers and the non-polar hydrophobic matrix. In this study, the effects of maleic anhydride-modified polyethylene additive, of carbon chains grafted by acetylation and of wood fiber contents on the tensile mechanical properties of the wood plastic composite were studied. Tensile tests were carried out using digital image correlation as an intrusiveness and robust method for strain measurements. Results showed first that the addition of wood fibers made the wood plastic composite stiffer but less flexible. Acetylation improved the interfacial adhesion properties: the Young modulus was increased and a lower strain at failure was reported. The coupling agent also increased the compatibility but mainly in the case where there was no grafted chain. With regard to the carbon chains, the number of grafts improved the elastic properties while their length did not appear to have any influence. Finally, a scanning electron microscope was used to characterize the post-mortem morphology of the fracture surfaces, the results of which supported the observations obtained from the tensile mechanical properties.

Cet article propose une methodologie d'inspection automatisee d'assemblages mecaniques basee sur l'utilisation d'un bras manipulateur, equipe en tete d'effecteur d'un capteur de vision artificielle. La methodologie d'inspection proposee dans cet article se base sur le couplage d'informations 2D et 3D, pour tirer profit de la rapidite de l'analyse en 2D et de la completude des donnees 3D.

A novel non-intrusive technique (stereo-correlation) was used to determine the apparent volume of a banana in convective drying condition. The volume was calculated using the 3D Digital Image Correlation method (3D-DIC), which provides the 3D shape of the banana during drying. The combination of this technique and mass measurement allows the calculation of the porosity using the model of Katekawa and Silva[ 1 ] and the graphical interpretation presented by Madiouli et al.[ 2 ] The banana shows an ideal shrinkage at the beginning of drying but stops shrinking at low moisture content, thus increasing the porosity up to 30–35%. The comparison of the experimental shrinkage and the calculated porosity with the experiments deduced from the literature enables us to conclude the effectiveness of the 3D-DIC technique as well as the porosity calculation model.

The tensile behavior of wood–plastic composite (WPC) with or without additive is studied using full-field strain measurements by 3D digital image correlation. It is shown that macroscopic values of the longitudinal strain are close to those measured by mechanical extensometer using standard mechanical tests. The modulus of elasticity provided by a Maxwell–Bingham model fitted to the experimental tensile curves is analyzed for several WPC formulations depending on the wood contents and the presence of additive. Color maps of the spatial strain distribution are commented. Moreover, the damage behavior and the degree of heterogeneity are analyzed thanks to the spatial standard deviation of the longitudinal strain field.

We propose a framework for obtaining synthetic speckle-pattern images based on successive transformations of Perlin's coherent noise function. In addition we show how a given displacement function can be used to produce deformed images, making this framework suitable for performance analysis of speckle-based displacement/strain measurement techniques, such as Digital Image Correlation, widely used in experimental mechanics.

A novel approach for correcting both spatial and drift distortions that are present in scanning electron microscope (SEM) images is described. Spatial distortion removal is performed using a methodology that employs a series of in-plane rigid body motions and a generated warping function. Drift distortion removal is performed using multiple, time-spaced images to extract the time-varying relative displacement field throughout the experiment. Results from numerical simulations clearly demonstrate that the correction procedures successfully remove both spatial and drift distortions. Specifically, in the absence of intensity noise the distortion removal methods consistently give excellent results with errors on the order of ±0.01 pixels. Results from the rigid body motion and tensile loading experiments at 200× indicate that, after correction for distortions, (a) the displacements have nearly random variability with a standard deviation of 0.02 pixels; (b) the measured strain fields are unbiased and in excellent agreement with previous full-field experimental data obtained with optical illumination; (c) the strain field variability is on the order of 60 microstrain in all components with a spatial resolution on the order of 25 pixels. Taken together, the analytical, computational and experimental studies clearly show that the correction procedures successfully remove both spatial and drift distortions while retaining excellent spatial resolution, confirming that the SEM-based method can be used for both micromaterial and nanomaterial characterization in either the elastic or elastic–plastic deformation regimes.