An official website of the United States government
Abstract In‐field visual inspections have inherent challenges associated with humans such as low accuracy, excessive cost and time, and safety. To overcome these barriers, researchers and industry leaders have developed image‐based methods for automatic structural crack detection. More recently, researchers have proposed using augmented reality (AR) to interface human visual inspection with automatic image‐based crack detection. However, to date, AR crack detection is limited because: (1) it is not available in real time and (2) it requires an external processing device. This paper describes a new AR methodology that addresses both problems enabling a standalone real‐time crack detection system for field inspection. A Canny algorithm is transformed into the single‐dimensional mathematical environment of the AR headset digital platform. Then, the algorithm is simplified based on the limited headset processing capacity toward lower processing time. The test of the AR crack‐detection method eliminates AR image‐processing dependence on external processors and has practical real‐time image‐processing.
more »
« less
This content will become publicly available on July 1, 2026
Increasing human immersion with image analysis using automatic region selection
Image-based models for defect quantification are fast and accurate but they are neither designed for real-time image processing in the field, nor do they incorporate humans in their decision-making process. Recently, researchers have integrated image-based inspection models for real-time defect quantification in Augmented Reality (AR) headsets to include human input in models’ decisions. However, deploying real-time image-based models in immersive devices is limited by their current minimal embedded processing capabilities. As a result, the model faces challenges with processing complexity timely, which limits human immersion during inspection using AR. To address this problem, this study introduces AR-ROI algorithm which integrates an automatic Region of Interest (ROI) selection method into an image-based defect quantification model and investigates the impact on processing time when deployed in an AR headset. This approach divides images into segments and initially processes all segments horizontally using the Canny algorithm until the number of positive pixels in a segment meets a threshold. The algorithm then vertically processes adjacent segments in subsequent row that both meet the threshold and are next to the segment from the previous row with the highest positive pixel count. This process continues iteratively and terminates when reaching a row without segments meeting the threshold or the final segment. Analytically, the algorithm reduces the asymptotic runtime by a factor of m’/m, where m and m’ are the pixel count in each row of an images and a segment, respectively. The results of this study are validated experimentally under various scenarios. The outcome of the experiments quantify the optimized processing time, while confirming the accuracy and analytical complexity assessment.
more »
« less
- Award ID(s):
- 2123346
- PAR ID:
- 10658842
- Publisher / Repository:
- Elsevier
- Date Published:
- Journal Name:
- Expert Systems with Applications
- Volume:
- 284
- Issue:
- C
- ISSN:
- 0957-4174
- Page Range / eLocation ID:
- 127938
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
This paper presents a pixel-level predictive sampling method for image sensing and processing to reduce the computing overhead for power-limited image sensing systems. The predictive sampling method scans through rows and columns to identify the location and value of the critical pixels, which are the turning points in the row and column arrays. The prediction is performed using the value of prior pixels and a pre-defined error threshold. When the prediction is successful, the pixel is marked as a non-critical pixel and is skipped for recording and processing. Only the critical pixels are selected for further processing. We proposed reconstruction methods that recover the raw image from the selected critical pixels using interpolation. The experimental results show that the proposed method can reduce the data throughput by 72% with an error of 1.6% for sparse images. The convolutional neural network model applied with this method can achieve a similar detection accuracy in a standard method while only using 27.1% of data size.more » « less
-
Purpose: Fluorescence imaging is a valuable tool for studying tear film dynamics andcorneal staining. Automating the quantification of fluorescence images is a challenging necessary step for making connections to mathematical models. A significant partof the challenge is identifying the region of interest, specifically the cornea, for collected data with widely varying characteristics.Methods: The gradient of pixel intensity at the cornea–sclera limbus is used as the objective of standard optimization to find a circle that best represents the cornea. Results of the optimization in one image are used as initial conditions in the next imageof a sequence. Additional initial conditions are chosen heuristically. The algorithm iscoded in open-source software.Results: The algorithm was first applied to 514 videos of 26 normal subjects, for a total of over 87,000 images. Only in 12 of the videos does the standard deviation in thedetected corneal radius exceed 1% of the image height, and only 3 exceeded 2%. The algorithm was applied to a sample of images from a second study with 142 dry-eye subjects. Significant staining was present in a substantial number of these images. Visual inspection and statistical analysis show good resuls for both normal and dry-eye images.Conclusion: The new algorithm is highly effective over a wide range of tear film andcorneal staining images collected at different times and locations.more » « less
-
Abstract The world’s coastlines are spatially highly variable, coupled-human-natural systems that comprise a nested hierarchy of component landforms, ecosystems, and human interventions, each interacting over a range of space and time scales. Understanding and predicting coastline dynamics necessitates frequent observation from imaging sensors on remote sensing platforms. Machine Learning models that carry out supervised (i.e., human-guided) pixel-based classification, or image segmentation, have transformative applications in spatio-temporal mapping of dynamic environments, including transient coastal landforms, sediments, habitats, waterbodies, and water flows. However, these models require large and well-documented training and testing datasets consisting of labeled imagery. We describe “Coast Train,” a multi-labeler dataset of orthomosaic and satellite images of coastal environments and corresponding labels. These data include imagery that are diverse in space and time, and contain 1.2 billion labeled pixels, representing over 3.6 million hectares. We use a human-in-the-loop tool especially designed for rapid and reproducible Earth surface image segmentation. Our approach permits image labeling by multiple labelers, in turn enabling quantification of pixel-level agreement over individual and collections of images.more » « less
-
Civil infrastructure inspection in hazardous areas such as underwater beams, bridge decks, etc., is a perilous task. In addition, other factors like labor intensity, time, etc. influence the inspection of infrastructures. Recent studies [11] represent that, an autonomous inspection of civil infrastructure can eradicate most of the problems stemming from manual inspection. In this paper, we address the problem of detecting cracks in the concrete surface. Most of the recent crack detection techniques use deep architecture. However, finding the exact location of crack efficiently has been a difficult problem recently. Therefore, a deep architecture is proposed in this paper, to identify the exact location of cracks. Our architecture labels each pixel as crack or non-crack, which eliminates the need for using any existing post-processing techniques in the current literature [5,11]. Moreover, acquiring enough data for learning is another challenge in concrete defect detection. According to previous studies, only 10% of an image contains edge pixels (in our case defected areas) [31]. We proposed a robust data augmentation technique to alleviate the need for collecting more crack image samples. The experimental results show that, with our method, significant accuracy can be obtained with very less sample of data. Our proposed method also outperforms the existing methods of concrete crack classification.more » « less
