다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

Hybrid pixel-level concrete crack segmentation and quantification across complex backgrounds using deep learning

Cracking is a common pavement distress that would cause further severe problems if not repaired timely, which means that it is important to accurately extract the information of pavement c...

Automated pavement crack detection and segmentation based on two‐step convolutional neural network

Road pavement cracks detection has been a hot research topic for quite a long time due to the practical importance of crack detection for road maintenance and traffic safety. Many methods have been proposed to solve this problem. This paper reviews the three major types of methods used in road cracks detection: image processing, machine learning and 3D imaging based methods. Image processing algorithms mainly include threshold segmentation, edge detection and region growing methods, which are used to process images and identify crack features. Crack detection based traditional machine learning methods such as neural network and support vector machine still relies on hand-crafted features using image processing techniques. Deep learning methods have fundamentally changed the way of crack detection and greatly improved the detection performance. In this work, we review and compare the deep learning neural networks proposed in crack detection in three ways, classification based, object detection based and segmentation based. We also cover the performance evaluation metrics and the performance of these methods on commonly-used benchmark datasets. With the maturity of 3D technology, crack detection using 3D data is a new line of research and application. We compare the three types of 3D data representations and study the corresponding performance of the deep neural networks for 3D object detection. Traditional and deep learning based crack detection methods using 3D data are also reviewed in detail.

/pdf/review-of-pavement-defect-detection-methods-3dj5w2pbs9.pdf

Review of Pavement Defect Detection Methods

Structural crack detection using deep convolutional neural networks

In this paper, we proposed an automatic crack detection method based on deep convolutional neural network −U-Net [4]. Unlike existing machine learning based crack detection methods, we can process an image as a whole without patchifying, thanks to the encoder-decoder structure of U-Net. The segmentation result is output from the network as a whole, instead of aggregation from neighborhood patches. In addition, a new cost function based on distance transform is introduced to assign pixel-level weight according to the minimal distance to the ground truth segmentation. In experiments, we test the proposed method on two datasets of road crack images. The pixel-level segmentation accuracy is above 92% which outperforms other state-of-the-art methods significantly.

Pixel-level Crack Detection using U-Net

Automatic welding of complex joints relies on scanning and inspection of the actual joint geometry, which is important for welding path planning and welding parameter setting. A robotic vision inspection system is presented in this paper to assess the geometry of complex joints before welding. The system can scan the joint with high accuracy and high resolutions via accurate system calibration and fast data synchronization. The scanned joint geometry is inspected to assess the quality of critical dimensions. Critical features are extracted for path planning, and the joint surface is segmented into different regions for assigning different welding parameters. Experimental results based on straight-joint and Y-joint are presented to show how the proposed system works.

Robotic Vision Inspection of Complex Joints for Automatic Welding

Visual inspection involves dimensional inspection and surface inspection. Dimensional inspection is usually achieved via comparing the scanned model of the inspection target with the CAD model to detect dimensional defects such as bending and missing material. Surface inspection is usually based on textures via image processing and analysis to detect defects such as pitting, crack, scratch, and corrosion. Existing systems used fixed predefined viewpoints to capture geometry and texture from the object. However, defects may appear at different locations, distinguishable defect features may not be captured by the fixed viewpoints. Thus, dynamical planning of the viewpoints is desired to provide better sensing parameters adaptive to actual local geometry, appearance and lighting conditions. This paper proposes a new robotic sensing framework for adaptive visual inspection. Viewpoints are planned based on individual object adaptive to different defect locations. Sensor parameters at each viewpoints are adaptive to the surface conditions for optimized geometry/image quality. Data from all viewpoints are fused together to provide an adaptive model.

Joint texture and geometry analysis for robotic adaptive visual inspection

Dimensional measurement is critical for quality control. Manual dimensional measurement using standard gauges can only be applied on a few datums. To measure a huge number of datums, a component needs to be scanned into a point cloud and measured digitally. For precision components, datum generation on the scanned point cloud is labor-intensive. Given a raw point cloud from scanner, this paper proposes an automatic dimensional measurement solution with an adaptive local registration algorithm and an adaptive datum generation algorithm. Using datums on the CAD model as reference, the adaptive local registration algorithm selects local regions on the scanned model to compensate the local deviation between the CAD model and the scanned model. After that, with outliers and noises in the raw data, the adaptive datum generation algorithm creates the correct datums on the scanned model adaptive to the actual geometry. Dimensional measurement based on the generated datums can be achieved automatically. Moreover, the solution does not require users to manually preprocess the point cloud, such as outlier and noise removal. As such, it improves the productivity in dimensional inspection.

Yusha Li

Papers

Pixel-level Crack Detection using U-Net

Robotic Vision Inspection of Complex Joints for Automatic Welding

Joint texture and geometry analysis for robotic adaptive visual inspection

Automatic Dimensional Measurement using Datums Generated from Point Clouds