Sub-pixel Edge Detection Research Articles

An online visual measurement method for workpiece dimension based on deep learning

Image preprocessing and edge detection are the two most critical steps when applying machine vision to measure workpiece dimension in industrial field. The surface of the workpiece is usually covered with interference regions, which makes edge detection difficult. Aiming at the problem, the Full Convolutional Network (FCN) model is used to detect interference regions of workpiece images, and then interference regions are processed by the directional texture repair method. To avoid the influence of workpiece surface texture after rough machining, a method based on Holistically-nested Edge Detection (HED) model for rough edge detection of workpiece images is proposed. The edge is obtained by HED, and then it is post-processed based on the non-maximum value suppression and double threshold segmentation methods in the Canny operator to obtain a refined edge image. In order to further improve the accuracy of the dimension measurement, sub-pixel level edge detection accuracy is achieved by cubic spline interpolation. Finally, the method is validated by a shaft workpiece. The measuring accuracy of the outer diameter of the workpiece can reach 0.02 mm, which can effectively meet the requirements of fast semi-finishing inspection. This research provides a common workpiece dimension measurement method in industrial field and methodological guidance for the application of deep learning in industrial inspection.

Research on O-ring Dimension Measurement Algorithm Based on Cubic Spline Interpolation

Current O-ring dimension measurement algorithms based on machine vision are mainly whole-pixel level algorithms, which have the disadvantage of a low measurement accuracy. In order to improve the stability and accuracy of O-ring dimension measurement, a sub-pixel edge detection algorithm based on cubic spline interpolation is proposed for O-ring dimension measurement. After image pre-processing of the O-ring graphics, the whole-pixel-level O-ring edges are obtained by using a noise-resistant mathematical morphology method, and then the sub-pixel edge contours are obtained using a sub-pixel edge detection algorithm based on cubic spline interpolation. Finally, the edge curve is fitted with the least squares method to obtain its inner and outer diameter as well as the size of the wire diameter. The experimental data show that the algorithm has a mean square error of 4.8 μm for the outer diameter and 0.18 μm for the wire diameter. The outer diameter error is kept within ±100 μm and the wire diameter error can be kept within ±15 μm. Compared with the whole pixel algorithm, the measurement accuracy has been greatly improved.

A Novel Subpixel Edge Detection Method of Pantograph Slide in Complicated Surroundings

Pantograph slide is a vulnerable component because of the contact and friction between the pantograph and catenary while the train runs with a high speed. Condition monitoring of the slide is a necessary and meaningful work to improve the operation efficiency and reduce the occurrence of accident. It is a hard work to measure the slide abrasion due to the dynamic and complicated environment. In this article, an accurate edge extraction method based on image analysis is proposed, which aims at providing an image metric property based on vision measurement to detect the wear of slide. First, a subpixel edge detection method is proposed. The gradient in horizontal direction is first taken into account, which filters out the nonhorizontal edges. Then, the upper and lower edges of the slide are located using the geometric restriction. The subpixel edge feature are also extracted in the image process, which can be used to the vision measurement system for monitoring the wear condition of the slide. The experiment results shows that the proposed algorithm can precisely detect the subpixel edges of the slide. The measurement error of the slide is less than 2 mm, which demonstrates that the proposed algorithm has a high accuracy in the large field of view.

Chisel edge wear measurement of high-speed steel twist drills based on machine vision

Tool wear is usually the most relevant parameter of tool performance detection, which directly affects the product quality and tool life. Online tool condition monitoring can effectively reduce workpiece scrap and machine failure. Chisel edge wear is one of the main wear forms in high-speed steel (HSS) twist drill. A machine vision method for chisel edge wear measurement is proposed in this paper to improve measure accuracy and reduce testing costs. Firstly, Otsu segmentation based on Laplacian edge information is proposed for tool image segmentation. A morphology-based Canny operator is developed to perform edge detection and image registration for rough positioning of the tool wear area. Zernike moment-based sub-pixel edge detection is proposed to improve the measurement accuracy. Finally, principal curve is employed to fit sub-pixel edge points to obtain smooth edge curve of the tool wear. Finally, the chisel edge wear can be detected and measured effectively based on the proposed method. In the real machining process, the test results show that the proposed system shows high response speed and inspecting accuracy, which is more convenient and efficient than manual measurement with an electron microscope. This system can be effectively applied to the real-time monitoring of tool wear in industry.

High-temperature thermal expansion behaviour of C/SiC studied using an in-situ optical visualisation method and numerical simulations in a quartz lamp array heating environment

The thermal expansion behaviour and thermal insulation performance of high-temperature structural ceramics have attracted considerable attention in the aerospace field. To further understand the high-temperature deformation behaviour and mechanism of C/SiC, an in-situ optical visualisation system was used to observe the variation in the gap width between C/SiC specimens in a quartz lamp array heating test. By using sub-pixel edge detection for the images recorded by the CCD camera and calculating the variation in gap width, it was found that the gap width first decreased and then increased during the heating process. Numerical simulations showed that the variation was caused by heat conduction and expansion of the base under the C/SiC specimens. The variation in the gap width under different working conditions was studied, and the thermal insulation performance of the specimens with and without thermal barrier coatings was compared. This work provides a new route with theoretical support for the study of the thermal expansion behaviour of high-temperature structural ceramics.

Concentric circle grids for camera calibration with considering lens distortion

The accuracy of camera calibration depends significantly on the control point detection which customarily subjects to image noise, distortion and resolution. This paper uses concentric circle grids as the calibration pattern and accordingly proposes a robust and accurate control point detection method for camera calibration. Each set of concentric circles on the grids pattern consists of multiple circles whose radii decrease progressively. The geometric structure of multi-hierarchies can not only robustly extract control points from complicated scenarios but also build a voting scheme to avoid numerical instability resulting from the estimation of projected centers. According to the common self-polar triangle of concentric circles and the local sub-pixel edge detection algorithm, we efficiently estimated sub-pixel coordinates of control points under projective transformation. Simulations and real experiments were conducted to demonstrate that the performance of calibration using concentric circle grids is superior to the popular grid patterns with respect to the image noise, distortion and resolution.

Research on Subpixel Algorithm of Fixed-Point Tool Path Measurement.

Tool safety is an important part of machining and machine tool safety, and machine tool path image detection can effectively obtain the in-machine condition of a tool. To obtain an accurate image edge and improve image processing accuracy, a novel subpixel edge detection method is proposed in this study. The precontour is segmented by binarization, the second derivative in the neighborhood of the demand point is calculated, and the obtained value is sampled according to the specified rules for curve fitting. The point whose curve ordinate is 0 is the subpixel position. The experiment proves that an improved subpixel edge can be obtained. Results show that the proposed method can extract a satisfactory subpixel contour, which is more accurate and reliable than the edge results obtained by several current pixel-level operators, such as the Canny operator, and can be used in edge detection with high-accuracy requirements, such as the contour detection of online tools.

Improved measure of workpiece surface deterioration during turning using non-contact vision method

The surface finish quality of a machined workpiece is commonly measured using the average roughness parameter, Ra. This parameter, however, is insensitive to the lateral changes undergone by the surface in the feed direction as a consequence of tool wear. In this work, the effectiveness of four methods of workpiece surface analysis, namely autocorrelation, cross-correlation, and two new methods, called lateral material shift (LMS) ratio and profile slope ratio (PSR) analyses are investigated. Dry machining experiments were carried out on 316 stainless steel. Images of tool nose and workpiece profiles were captured using digital camera, and the edges were extracted using sub-pixel edge detection. In the autocorrelation approach, each workpiece profile was correlated with a shifted version of the same profile. In the cross-correlation approach, the workpiece profiles at different stages of machining were correlated with a reference profile generated using the unworn tool edge. In the LMS ratio method, the material shift ratios were determined from each waveform on the workpiece profile at various stages of tool wear, while in the PSR method the slopes at the right and left part of the waveform were compared. Among the four methods, the LMS ratio method produced the best correlation with tool flank wear with the maximum R-squared value of 0.9461, while average roughness Ra showed no correlation at all with both major and nose flank wear.

Effect of Vibration on Surface Roughness in Finish Turning: Simulation Study

This paper presents the results of simulation analyses carried out to investigate the effect of adding controlled vibration of varying amplitudes and frequencies onto simulated workpiece profiles on surface roughness. The surface profiles were generated using the nose edge image of a real cutting tool insert at various stages of wear. The invariant moment sub-pixel edge detection method was used to extract the precise tool nose profile. The extracted nose profile was duplicated 20 times to generate the simulated workpiece profile. Vibration signals in the form of sine waves of amplitudes 0 to 50 % of the peak-to-valley height of the original workpiece profile and various frequencies were added to the simulated profile. Noise signals emanating from real machining were also added to the original profile to investigate their effect on roughness. The simulation study has shown that average roughness may increase continuously or fluctuate randomly depending on the magnitude of vibration added to the vibration-free workpiece profile. In the absence of vibration, the surface roughness of the workpiece decreases due to flattening effect of the tool nose during early stages of tool wear.

Monocular Vision-Based Calibration Method for Determining Frequency Characteristics of Low-Frequency Vibration Sensors

Calibration is required to determine the frequency characteristics of vibration sensors to ensure their measurement accuracy in engineering applications. Thus, a monocular vision-based calibration method for low-frequency vibration sensors is investigated in this study. A sub-pixel edge detection method which based on Gaussian curve fitting is applied to extract the edges of motion sequence images in order to accurately measure the excitation acceleration of the sensors. Because the motion sequence images and the output signal of the sensors cannot be collected synchronously, it is very difficult to align the excitation acceleration signal obtained from the images and the output signal in the time domain. Although the misalignment only has a negligible effect on the magnitude frequency characteristic calibration, it dramatically decreases the calibration accuracy of the phase frequency characteristic, especially with the increasing of the frequency. A time-spatial synchronization technique is proposed to accurately calibrate the phase frequency characteristic by determining the phase of the excitation acceleration signal at a specific spatial position and that of the output signal at the corresponding time. Finally, both the magnitude and phase frequency characteristics are simultaneously calibrated by the investigated method with a flexible and low-cost vision system. The experimental results compared with laser interferometry show that the investigated method accomplishes the high-accuracy magnitude and phase frequency characteristics calibration in a broad low-frequency range. Its calibration accuracy is superior to that of laser interferometry when the frequency is less than 0.3 Hz, and is equivalent at other frequencies.

Vision Positioning method for Autonomous Precise Landing of UAV Based on Square Landing Mark

Rotor-craft is a kind of VTOL UAV and is widely used in multiple fields. Among relative researches, vision guided autonomous landing of rotor-craft has been a hot spot, where vision positioning is the most crucial. The core of the algorithm is to calculate the position and attitude information according to the change of the visual image of the same object at different time or frame. Based on the real-time self calibration technology of airborne monocular vision, this paper puts forward a method that takes the designed landing mark composed of black and white squares as the cooperative target, solving the relative pose of camera and cooperative target to carry out the landing positioning of UAV, and realizes the full-automatic sub-pixel precision linear edge detection to ensure the accuracy of visual positioning. A series of simulation experiments show that for 768 * 576 pixel image, when the camera is about 12m away from the target and the noise deviation reaches 3 pixels, the total time of edge extraction and calibration is 0.511s, and the position and attitude estimation are still satisfactory, which shows that the proposed algorithm can effectively realize the autonomous landing of UAV.

Adaptive Image Edge Extraction Based on Discrete Algorithm and Classical Canny Operator

In order to improve the accuracy of image edge detection, this paper studies the adaptive image edge detection technology based on discrete algorithm and classical Canny operator. First, the traditional sub-pixel edge detection method is illustrated based on the related literature research. Then, Canny operator is used for detection, the edge model of the quadric curve is established using discrete data, and the adaptive image edge parameters are obtained using one-dimensional gray moment. Experimental results show that the accuracy of feature detection is 99%, which can be applied to the practice of image edge detection to a certain extent.

Deep learning-based super-resolution images for synchronous measurement of temperature and deformation at elevated temperature

Measuring temperature and deformation at elevated temperature has recently been a major concern for the engineering tests and the quality improvement of captured images is critical to it. Here, we propose a simple, high-precision and easy-to-implement technique combing the image capturing and processing methods to obtain the high-resolution images as well as the corresponding temperature and deformation fields. The super-resolution convolutional neural network (SRCNN) algorithm is used for the super-resolution reconstruction, while the deformation field is calculated by the DIC method and the temperature field is synchronous obtained by the improved two-color method. Flame heating experiment of the SiC material validated the applicability and the improvement of the proposed method, showing the improvement in image quality (with PSNR increased 16.7 %) and calculation accuracy of temperature and deformation (with error decreased 3.57 %). Meanwhile, the phenomenon of "feature point drift" is pointed out and the error of deformation measurement for high-resolution and low-resolution is analyzed. Furthermore, the sub-pixel edge detection can be achieved by the combination of the proposed method and the edge detection method, which shows a potential value in detecting of surface defects or cracks.

Applying Color Doppler Image based Virtual Surgery in Placenta Previa Cesarean Section

Abstract Objective: To study the application of image processing technology in cesarean section of placenta previa, thereby reducing the occurrence of high-risk pregnancy. Methods: First, the method of gray image enhancement is analyzed. This method enhances the gray difference between the target and the background, highlights useful information, summarizes the source and type of noise, and proposes common filtering and noise reduction methods to suppress the noise. For edge detection, pixel-level edge detection operators and sub-pixel-level edge detection operators are summarized. The Canny edge detection operator and the Gaussian fitting sub-pixel edge detection operator are introduced in detail, and innovative improvements are carried out for resolving the deficiencies of the algorithm. Results: The improved adaptive iterative segmentation thresholding method results in a threshold of T = 98 and 11 iterations. The image segmentation quality of the improved Otsu method has been greatly enhanced. After the second segmentation, the improved Otsu method finds the optimal threshold T = 76. Conclusion: Color Doppler ultrasound image processing technology has excellent application in placenta previa cesarean section.

Automatic Wear Measurement of Pantograph Slider Based on Multiview Analysis

Pantograph slider is a critical component of the electrical equipment to obtain power sources for the electric locomotive, whereas the interaction between the slider and the overhead contact line can cause persistent surface wear of the collector strip. The condition monitoring for the level of abrasion of the pantograph slider is significant to avoid major damage. In this article, an innovative approach based on multiview analysis is proposed for monitoring the surface wear of the slider in a running train, accurately measuring the residual thickness of the slider, and locating the wear defects appearing. Especially, the 3-D reconstruction technology is used for detecting the wear quantity between the upper and lower edges of the slider. First, the target edges of the slider are determined using the proposed subpixel edge detection algorithm. Second, mutual matching of the edge points between the multiview images is finished using the epipolar geometric constraint. Moreover, the 3-D reconstructions of the matching points are computed with the principle of triangulation. Finally, the wear volume of the slider can be determined. Experimental results show that the proposed method can successfully measure the abrasion of the pantograph slider under complex conditions, which demonstrate the applicability, effectiveness, and good performance of the proposed approach.

Active Contour Method Based Sub-pixel Critical Dimension Measurement of Thin Film Transistor Liquid Crystal Display (TFT-LCD) Patterns

A novel application-oriented sub-pixel measurement method utilizing an active contour model with image resampling is presented for a critical dimension (CD) measurement of pixel patterns of a flat-panel thin film transistor crystal display (TFT-LCD) and organic light-emitting diode display. In modern manufacturing of flat panel display, optical measurement of each dimension of pixel patterns has a critical role in the process control, and the resolution of a measurement system becomes a limitation of the manufacturing process. Several methods have been developed to overcome this limit at the image post processing stage, but as the level of pixel integration and manufacturing throughput are increasing rapidly, more robust and effective inspection approach is required. In this paper, a novel sub-pixel level edge detection algorithm with active contour method and fast pixel resampling is proposed for micron scale CD measurement, and its advantages on measurement repeatability and noise handling are presented, along with the results of various industrial sample measurements and comparison with conventional critical dimension measurement algorithms.

High-Precision Vision Sensor Method for Dam Surface Displacement Measurement

Dam surface displacement measurement can monitor the safety of dam operation and provide early warning for dam accidents. This paper proposes a method for measuring the displacement of the dam surface based on double vision sensors. Mathematical morphology is combined with the Roberts edge detection operator to complete the sub-pixel extraction of the marker edge. The method makes the edge of the marker in the image smoother compared with the Zernike moment sub-pixel edge detection algorithm. The edge gray center of gravity method is proposed to realize the positioning of the marker, which improves the precision of the vision measurement method. Additionally, this paper proposes the design of a test system and describes the experiments conducted in this study. The results revealed that the proposed method can achieve millimeter-level measurement accuracy and has the advantages of simple operation, short measurement period, online monitoring, and low cost, which satisfies the requirements for measuring the dam surface displacement.

High Precision Dimensional Measurement with Convolutional Neural Network and Bi-Directional Long Short-Term Memory (LSTM).

In modern industries, high precision dimensional measurement plays a pivotal role in product inspection and sub-pixel edge detection is the core algorithm. Traditional interpolation and moment methods have achieved some success. However, those methods still have shortcomings. For example, the accuracy is still insufficient with the resolution limitation of the image sensor. Moreover, prediction results can be affected by image noise. With the recent success of deep learning technology, we propose a sub-pixel edge detection method based on convolution neural network (CNN) and bi-directional long short-term memory (LSTM). First, one-dimensional visual geometry group-16 (VGG-16) is employed to extract edge features. Then, a transformation operation is developed to generate sequence information. Lastly, bi-directional LSTM with fully-connected layers is introduced to output edge positions. Experimental results on our steel plate dataset demonstrate that our method achieves superior accuracy and anti-noise ability than traditional methods.

A novel optical method for measuring 3D full-field strain deformation in geotechnical tri-axial testing

Soil is a complex multiphase material and has anisotropic properties; meanwhile, soil behaviour is typically evaluated in tri-axial testing. Until now, there have still been some challenges in measuring the total strain, the local strain deformation and the volume changes during tri-axial testing. In conventional geotechnical testing, it cannot acquire enough deformation characteristics. To improve the situation, a digital image measurement system has been constructed and calibrated on a geotechnical tri-axial apparatus. In the digital image measurement system, the three-dimensional digital image correlation (3D-DIC) technique was applied to measure the soil specimen’s 3D large deformation and reconstruct the 3D surface topography of the specimen. To improve the measurement accuracy, a standard spatial calibration procedure to rectify the 3D reconstruction data in tri-axial testing is proposed to improve the effect of the projective transformation in stereovision. Through rectification, the root-mean-squared error in terms of displacement measurement in the tri-axial testing was reduced by more than half. Furthermore, a subpixel edge detector was applied to estimate the volume changes based on the 3D-DIC results. Finally, comparing the results between the image measurement system and the conventional measurement method, the proposed method not only can obtain the specimen’s distribution of 3D full-field deformation, volume strain and so on, but also can provide a method for rectifying the 3D data measured by the image measurement technique in tri-axial testing. The results from 3D reconstruction can describe directly the characteristics of anisotropy and non-uniformity in geotechnical materials. The results can provide more information than conventional testing when establishing a constitutive model in soil mechanics.

A High-Reliable and High-Precision Algorithm of Angle Measurement for UAV Airborne Photoelectrical Detection System

In order to achieve high-reliable and high-precision target measurement for UAV photoelectric detection system without increasing any costs, an algorithm of angle measurement is proposed. Firstly, by analyzing the character of grating grid image’s edge, a fast method using three gradient points to establish a parabolic function is applied for automatic grating image sub-pixel edge detection. Coarse code decoding is improved by OTSU’s method to reduce bit error rate (BER) as well as manufacturing difficulty, and a high-precision subdivision method based on observed image of grating lines is introduced to calculate subdivision angle. Secondly, simulation of BER and precision analysis show that proposed algorithm is robust in both processes of coarse code decoding and subdivision. Finally, experiment results show that BER is no more than 0.02%, the standard deviation of proposed algorithm is 1.8″ and the standard deviation of error is only 7.1″ with one reading head, indicating that proposed algorithm is robust, reliable and precise.