A hybrid method for ellipse detection in industrial images

Xie

et al. 2018

Sensors

A handheld 3D laser scanning system is proposed for measuring large-sized objects on site. This system is mainly composed of two CCD cameras and a line laser projector, in which the two CCD cameras constitute a binocular stereo vision system to locate the scanner’s position in the fixed workpiece coordinate system online, meanwhile the left CCD camera and the laser line projector constitute a structured light system to get the laser lines modulated by the workpiece features. The marked points and laser line are both obtained in the coordinate system of the left camera in each moment. To get the workpiece outline, the handheld scanner’s position is evaluated online by matching up the marked points got by the binocular stereo vision system and those in the workpiece coordinate system measured by a TRITOP system beforehand; then the laser line with workpiece’s features got at this moment is transformed into the fixed workpiece coordinate system. Finally, the 3D information composed by the laser lines can be reconstructed in the workpiece coordinate system. A ball arm with two standard balls, which is placed on a glass plate with many marked points randomly stuck on, is measured to test the system accuracy. The distance errors between the two balls are within ±0.05 mm, the radius errors of the two balls are all within ±0.04 mm, the distance errors from the scatter points to the fitted sphere are distributed evenly, within ±0.25 mm, without accumulated errors. Measurement results of two typical workpieces show that the system can measure large-sized objects completely with acceptable accuracy and have the advantage of avoiding some deficiencies, such as sheltering and limited measuring range.

“…The 2D coordinates of the marked points are extracted from these 15 pairs of images with sub-pixel precision [ 17 ]. The unknown parameters in Equations (1)–(4), including

and

, are then calibrated by adopting the binocular stereo calibration function of OpenCV.…”

Section: Modeling and Calibration Of The Handheld 3d Laser Scanninmentioning

confidence: 99%

Research on a Handheld 3D Laser Scanning System for Measuring Large-Sized Objects

Xie

et al. 2018

Sensors

“…Over the past years, many ellipse detection algorithms sprang up and were studied broadly to solve the ellipse detection problem, which can be briefly grouped into several categories: Hough transform method [21][22][23][24], leastsquare fitting method [25][26][27][28][29], clustering method [30][31][32][33][34]. Hough transform has been widely used for detecting geometric primitives such as line segment, circle and ellipse.…”

Section: Introductionmentioning

confidence: 99%

“…Considering the images with strong noise, Lu [22] proposed an iterative randomized Hough transform method to detect the ellipses. Chen [23] developed a novel ellipse detection method based on Hough transform and edge fo llo wing method to save the memo ry and computation time. Co mb ined with artificial neural network, Serra [24] also presented a modified Randomized Hough Transform method to make mult iple scans over the acquired images.…”

Section: Introductionmentioning

confidence: 99%

Dynamic Granularity Matrix Space Model Based High Robust Multi-Ellipse Center Extraction Method for Camera Calibration

Xue

2020

IEEE Access

Ellipse center extract ion is the important basis of camera calib ration with circle arrays calibrat ion board, wh ich direct ly affects the measurement accuracy of mach ine vision system. A iming at the problem of high noise sensitivity of the most ellipse center extract ion methods, a high robust multi-ellipse center extraction method is proposed. The idea is to transform the problem of mult i-ellipse center extraction into the problems of ellipse sub-pixel edge segmentation and edge based ellipse center clustering. Firstly, the flowchart of the proposed method is introduced. Then, the theory of fuzzy quotient space is introduced to establish the dynamic granularity matrix space model, which describes image segmentation problem as the jump ing and transformat ion of the image by hierarchical structure. Finally, an adaptive multi-ellipse sub-pixel edge segmentation method and an optimal ellipse center ext raction method are proposed based on the dynamic granularity matrix space model. Experimental results sh ow that the proposed method has higher accuracy, robustness and lower camera calibration error. INDEX TERMS multi-ellipse center, dynamic granularity matrix space, sub-pixel edge segmentation, hierarchical structure, camera calibration.

“…Ellipse fitting arises in biometrics, exemplified by the application to iris segmentation and localization [ 13 – 15 ] face detection [ 16 ] and pathological brain detection [ 17 ]. Finally, ellipse fitting arises in the application to industrial inspection [ 18 ] and control of silicon single crystal growth [ 19 , 20 ].…”

Section: Introductionmentioning

confidence: 99%

“…A geometric distance measure featured also in a maximum likelihood estimation algorithm in [ 27 , 28 ]. Other measures include treating ellipse data points as a noisy signal and applying filtering techniques [ 29 ], use of Gaussian Mixture Models [ 30 ], projective invariants [ 31 ] and hybrid approaches [ 18 , 32 ]. In yet another approach, Kanatani and Rangarajan proposed hyperaccurate methods of ellipse fitting in [ 33 ] and [ 34 ], while Yu et al in [ 35 ] proposed a new distance metric based on some intrinsic properties of ellipses and spheroids.…”

Section: Introductionmentioning

confidence: 99%

Pre-processing by data augmentation for improved ellipse fitting

2018

Ellipse fitting is a highly researched and mature topic. Surprisingly, however, no existing method has thus far considered the data point eccentricity in its ellipse fitting procedure. Here, we introduce the concept of eccentricity of a data point, in analogy with the idea of ellipse eccentricity. We then show empirically that, irrespective of ellipse fitting method used, the root mean square error (RMSE) of a fit increases with the eccentricity of the data point set. The main contribution of the paper is based on the hypothesis that if the data point set were pre-processed to strategically add additional data points in regions of high eccentricity, then the quality of a fit could be improved. Conditional validity of this hypothesis is demonstrated mathematically using a model scenario. Based on this confirmation we propose an algorithm that pre-processes the data so that data points with high eccentricity are replicated. The improvement of ellipse fitting is then demonstrated empirically in real-world application of 3D reconstruction of a plant root system for phenotypic analysis. The degree of improvement for different underlying ellipse fitting methods as a function of data noise level is also analysed. We show that almost every method tested, irrespective of whether it minimizes algebraic error or geometric error, shows improvement in the fit following data augmentation using the proposed pre-processing algorithm.