Search

Taiwan, frequently affected by extreme weather causing phenomena such as earthquakes and typhoons, faces a high incidence of rockfall disasters due to its largely mountainous terrain. These disasters have led to numerous casualties, government compensation cases, and significant transportation safety impacts. According to the National Science and Technology Center for Disaster Reduction records from 2010 to 2022, 421 out of 866 soil and rock disasters occurred in eastern Taiwan, causing traffic disruptions due to rockfalls. Since traditional sensors of disaster detectors only record changes after a rockfall, there is no system in place to detect rockfalls as they occur. To combat this, a rockfall detection and tracking system using deep learning and image processing technology was developed. This system includes a real-time image tracking and recognition system that integrates YOLO and image processing technology. It was trained on a self-collected dataset of 2490 high-resolution RGB images. The system's performance was evaluated on 30 videos featuring various rockfall scenarios. It achieved a mean Average Precision (mAP50) of 0.845 and mAP50-95 of 0.41, with a processing time of 125 ms. Tested on advanced hardware, the system proves effective in quickly tracking and identifying hazardous rockfalls, offering a significant advancement in disaster management and prevention. [ABSTRACT FROM AUTHOR]

Simple Summary: We proposed a deep learning algorithm to detect lymph nodes and classify them in the head and neck region on computed tomography. We further analyzed the inference result from the model and found that the size of the lymph nodes may be a characteristic for the model to classify them. This finding is consistent with current clinical aspects. We will deploy the model in clinical practice and hope to assist clinicians in finding out the lesions more correctly and efficiently. Background: Head and neck cancer is highly prevalent in Taiwan. Its treatment mainly relies on clinical staging, usually diagnosed from images. A major part of the diagnosis is whether lymph nodes are involved in the tumor. We present an algorithm for analyzing clinical images that integrates a deep learning model with image processing and attempt to analyze the features it uses to classify lymph nodes. Methods: We retrospectively collected pretreatment computed tomography images and surgery pathological reports for 271 patients diagnosed with, and subsequently treated for, naïve oral cavity, oropharynx, hypopharynx, and larynx cancer between 2008 and 2018. We chose a 3D UNet model trained for semantic segmentation, which was evaluated for inference in a test dataset of 29 patients. Results: We annotated 2527 lymph nodes. The detection rate of all lymph nodes was 80%, and Dice score was 0.71. The model has a better detection rate at larger lymph nodes. For those identified lymph nodes, we found a trend where the shorter the short axis, the more negative the lymph nodes. This is consistent with clinical observations. Conclusions: The model showed a convincible lymph node detection on clinical images. We will evaluate and further improve the model in collaboration with clinical physicians. [ABSTRACT FROM AUTHOR]

Rice is one of the most significant crops cultivated in Asian countries. In Taiwan, almost half of the arable land is used for growing rice. The life cycle of paddy rice can be divided into several stages: vegetative stage, reproductive stage, and ripening stage. These three main stages can be divided into more detailed stages. However, the transitions between stages are challenging to observe and determine, so experience is required. Thus, rice cultivation is challenging for inexperienced growers, even with the standard of procedure (SOP) provided. Additionally, aging and labor issues have had an impact on agriculture. Furthermore, smart farming has been growing rapidly in recent years and has improved agriculture in many ways. To lower the entry requirements and help novices better understand, we proposed a random forest (RF)-based machine learning (ML) classification model for rice growth stages. The experimental setup installed in the experiment fields consists of an HD smart camera (Speed-dome) to collect the image and video data, along with other internet of things (IoT) devices such as 7-in-1 soil sensors, a weather monitoring station, flow meter, and milometer connected with LoRa base station for numerical data. Then, different image processing techniques such as object detection, object classification, instance segmentation, excess green index (EGI), and modified excess green index (EGI) were used to calculate the paddy height and canopy cover (CC) or green coverage (GC). The proposed ML model uses these values as input. Furthermore, growth-related factors such as height, CC, accumulative temperature, and DAT are used to develop our model. An agronomist has been consulted to label the collected different stages of data. The developed optimal model has achieved an accuracy of 0.98772, and a macro F1-score of 0.98653. Thus, the developed model produces high-performance accuracy and can be employed in real-world scenarios. [ABSTRACT FROM AUTHOR]

Defect pattern detection and classification are challenging for thin-film-transistor liquid-crystal display (TFT-LCD) manufacturing. Limitations of the existing solutions for automatic optical inspection can be traced in part to the lack of a framework within which different existing and new defect patterns can be analyzed, while integrating domain knowledge and effective technologies. This study aims to develop a framework for image-based defect classification that employs the convolution neural networks without using complex and time-consuming image-processing processes in advance. An empirical study was conducted in a leading TFT-LCD manufacturing in Taiwan for validation. The results have shown that the defect patterns can be effectively classified by the proposed convolutional neural networks that outperform the existing approaches such as Support Vector Machine and Random Forest. The developed solution is implemented to effectively support the engineers. [ABSTRACT FROM AUTHOR]

On modern industrial product lines, assembly is predominantly performed through manual labor or using semi-automatic equipment. With the development of automation, robots are becoming crucial in improving the flexibility of production processes. This paper presents an assembly method that employs image processing to control industrial robots in solving a sphere assembly puzzle. By using image-processing techniques—such as block recognition, circle detection, and camera calibration—the system can detect and move randomly placed blocks to the correct assembly positions. Cooperation is also integrated into the system, allowing operators to intervene in the assembly and improving the system's efficiency while ensuring safety of operators. The results of the application on a practical platform using a HIWIN robot RT605-GB reveal a high rate of successful assembly, providing evidence that the system is effective. [ABSTRACT FROM AUTHOR]

On 6 February 2018 at 23:50 local time, a Mw 6.4 earthquake struck eastern Taiwan. We characterize the instantaneous surface ground motion and the permanent displacement induced by this event from continuous GPS data and SAR images within a short time after the mainshock. We use high-rate GPS positioning techniques to obtain epoch-by-epoch positions peak ground displacement to assess potential seismic damage. The maximum coseismic GPS horizontal displacement of about 450 mm trending to the northeast is observed at the station HUAL located on the hanging wall of the Milun fault. The PEPU located on the footwall of the Milun fault shows a coseismic horizontal displacement of 280 mm trending to the southwest and a coseismic uplift of about 70 mm. Moreover, ascending and descending tracks of ALOS-2 and Sentinel-1 SAR images are processed to estimate coseismic surface deformation along the line-of-sight (LOS) toward the satellite. Then, wide coverage from east-west and uplift components of surface deformation is fulfilled by combining the LOS displacement from ascending and descending interferograms. The main deformation area revealed by both GPS results and D-InSAR interferograms is concentrated around the Milun and Lingding faults. Significant uplift on the footwall of the northern Lingding Fault implies that the Milun fault and an unknown westdipping fault close to the Lingding fault were triggered. Both the two nodal planes of the Mw 6.4 Hualien event could be different with the kinematic behavior of the Milun fault and Lingding fault. Thus we suggest that slip on multiple faults was triggered during the 0206 event. [ABSTRACT FROM AUTHOR]

Background: With the increasing prevalence of Picture Archiving and Communication Systems (PACS) in healthcare institutions, there is a growing need to measure their success. However, there is a lack of published literature emphasizing the technical and social factors underlying a successful PACS. Methods: An updated Information Systems Success Model was utilized by radiology technologists (RTs) to evaluate the success of PACS at a large medical center in Taiwan. A survey, consisting of 109 questionnaires, was analyzed by Structural Equation Modeling. Results: Socio-technical factors (including system quality, information quality, service quality, perceived usefulness, user satisfaction, and PACS dependence) were proven to be effective measures of PACS success. Although the relationship between service quality and perceived usefulness was not significant, other proposed relationships amongst the six measurement parameters of success were all confirmed. Conclusions: Managers have an obligation to improve the attributes of PACS. At the onset of its deployment, RTs will have formed their own subjective opinions with regards to its quality (system quality, information quality, and service quality). As these personal concepts are either refuted or reinforced based on personal experiences, RTs will become either satisfied or dissatisfied with PACS, based on their perception of its usefulness or lack of usefulness. A satisfied RT may play a pivotal role in the implementation of PACS in the future. [ABSTRACT FROM AUTHOR]

Some new 3D image technologies including image capturing, digital data processing, 2D/3D video conversion, and naked eye 2D/3D compatible display have been developed in Taiwan for the future 3D world applications. An innovative 3D image capturing and displaying architecture is a cost effective system for 3D content generation, transmission, and display. The multiview images of fine art such as sculpture can be captured by using 3D scanner or camera. The obtained multiview images can be transmitted though popular streaming channel to the cloud computing center that handles huge tasks of 2D/3D image processing. Then, the multi-view images are fed into the new types of 2D/3D compatible display. This new type of display using a specially designed microretarder can present 2D and naked eye 3D images on the same display simultaneously. Some new 3D applications with real-time 2D to 3D conversion technology are also introduced. [ABSTRACT FROM AUTHOR]

By analysing teleseismic waveform data, we propose a large-scale Taiwan Reference Moho Discontinuity Model (TRMDM) through coherent stacking and H–κ analysis of receiver functions (RFs). Tectonic structural features are determined from accurate crustal thickness ( H) and reliable Vp/ Vs ratio (κ) (consequently Poisson's ratio, σ) estimations for Taiwan. The best estimated regional average Moho depth is 33.5 ± 2.9 km. Noted is a westward crustal thinning from 39 ± 3.0 km in eastern Taiwan to 30 ± 1.5 km in southeastern China. The averaged Moho depth is 23.0 ± 2.5 km in northern Taiwan and increases to 33.5 ± 2.0 km in the south. The markedly thin crust and high Vp/ Vs ratio in northern Taiwan are mainly associated with backarc opening and magmatic activities. Based on H–κ estimations and associated seismicity patterns beneath Suang-Long (SSLB) and adjacent four stations, we propose two Moho reference models that exhibit distinctive crustal thickening in central Taiwan. The spatial variations of crustal thickness models show strong lateral change in the Moho topography. The thick crust in the south, thin crust in the north and transition in central Taiwan mark the different tectonic units between the subduction wedge and the collision prism onshore. In view of the following list of constraints: thicker crust in the south than in the north; eastward crust thickening; significant Moho depth changes; clear 9–11 s subcrustal boundaries from five stations; the compelling thickening of crust and upper mantle beneath the central mountain belt; seismicity patterns with the absence of mantle earthquake in the Wadati-Benioff zone; regional geological features; the occurrence of the inland Chi-Chi earthquake; the compelling low κ at YULB (1.58 ± 0.13) that may extend to central Taiwan at SSLB (1.59 ± 0.23); relatively thick crust at YULB (39.5 ± 1.6 km) and SSLB (40.5 ± 1.3 km or 52.5 ± 1.5 km), and finally, the apparent horizontal offset between highest Yu Shan Mountain and the deepest Moho at SSLB, we are able to suggest that the initial subduction of the Eurasian Plate occurred in southern Taiwan and most likely extended to the transition zone in central Taiwan. Although the current RF study is limited by large interstation distance, our analysis provides more notable constraints than previous studies to explicitly determine the characteristic features of Moho discontinuity. This detailed approach means that rather than treating the tectonic framework as a whole heterogeneous lithospheric feature, clear Moho boundary geometry variations indicate that active orogenic processes and differential tectonic stress loading create separate and complex crustal and subcrustal-to-upper-mantle structural features caused by the arc–continent collision. [ABSTRACT FROM AUTHOR]

Abstract: We use the two-dimensional (2D) electrical resistivity imaging (ERI) method to detect the contact between sediment and bedrock in a modern fluvial system. We performed eight 2D electrical resistivity surveys along the Peikang River in Kuohsing area, central Taiwan, with a pole–pole array of 1-m electrode spacing to produce a spatial resolution of about 1 m. The resistivity data were inverted into subsurface electrical structures using the least-squares inversion techniques. Then, the contact between the bedrock and sediment was found by using an image processing technique called Laplacian edge detection (LED), which represents an objective approach in interpreting various geophysical images. Numerical modeling demonstrates that isoresistivity lines are not in good agreement with the hypothetical interfaces between the bedrock and sediment, but the zero-lines of the Laplacian operation are strikingly accurate. Our results indicate a well-defined boundary in the resistivity structure that can be used to estimate the quantity of sediments covering bedrock, thus highlighting the utility of this technique in studies of landscape evolution, sediment transport, and sediment budgets. [Copyright &y& Elsevier]

Abstract: This work investigates atmospheric visibility by employing two statistical processes for high-frequency extraction: Sobel operator and fast Fourier transform (FFT). In replacing the traditional measurement methods for atmospheric visibility, the new approaches can provide accurate digital data urban visibility by establishing the numerical indices. The procedure is illustrated as follows. Digital images of urban areas in Kaohsiung, a city at the south of Taiwan, are analyzed according to brightness. High-frequency components of the image are extracted to calculate the index values by employing the Sobel operator and FFT. Finally, the correlation between each index and the visual range estimated by trained investigators are evaluated. A good correlation between two indices and the values obtained by visual investigation is affirmed by correlation coefficients (), 0.8139 and 0.7797, respectively. Furthermore, both indices are highly correlated with each other (). Convenient transmission and the exchangeability of digital images of the real-time landscape facilitate the publication of results on the world wide web (WWW). [Copyright &y& Elsevier]

In recent years, Taiwan's government has focused on policies regarding offshore wind farming near the Indo-Pacific humpback dolphin habitat, where marine mammal observation is a critical consideration. The present research developed an algorithm called National Taiwan University Passive Acoustic Monitoring (NTU_PAM) to assist marine mammal observers (MMOs). The algorithm performs whistle detection processing and whistle localization. Whistle detection processing is based on image processing and whistle feature extraction; whistle localization is based on the time difference of arrival (TDOA) method. To test the whistle detection performance, we used the same data to compare NTU_PAM and the widely used software PAMGuard. To test whistle localization, we designed a real field experiment where a sound source projected simulated whistles, which were then recorded by several hydrophone stations. The data were analyzed to locate the moving path of the source. The results show that localization accuracy was higher when the sound source position was in the detection region composed of hydrophone stations. This paper provides a method for MMOs to conveniently observe the migration path and population dynamics of cetaceans without ecological disturbance. [ABSTRACT FROM AUTHOR]

Coastal upwelling is important for coastal ecosystems and the blue economy because of its large productivity and large potential for catching fish. However, coastal upwelling along the Taiwan east coast has received little attention from the research community. This study used five-year daily Himawari-8 geostationary satellite sea surface temperature data to map the coastal upwelling east of Taiwan during the summer monsoon season. We applied a semi-automatic image process technique based on the topographic position index for the quantitative upwelling mapping. The results show clear evidence of seasonal coastal upwelling along the entire Taiwan east coast, mainly under the influence of upwelling-favorable southwesterly/southerly winds. There are three broad upwelling centers along the Taiwan east coast: north, central, and south. The upwelling around the northern center has the longest upwelling season, lasting from May to September. The upwelling extents are larger between June and August during the height of the summer monsoon. [ABSTRACT FROM AUTHOR]

Detecting road damage quickly and accurately facilitates the ability of road-maintenance agencies to make timely repairs to road surfaces, maintain optimal road conditions, optimize transportation safety, and minimize transportation costs. An extensive evaluation of eight deep-learning-based road-damage detection models was conducted in this study. Each model was trained on 9493 images sourced from multiple databases. The 16165 instances of road damage in these images were categorized into five types of damage, including longitudinal crack, horizontal crack, alligator damage, pothole-related crack, and line blurring. Two experiments were conducted that identified two models, single shot multi-box detector (SSD) Inception V2 and faster region-based convolutional neural networks (R-CNN) Inception V2, as providing the best balance of road-damage-detection accuracy and image processing time. These experiments demonstrated that increasing the diversity of image sources improved road-damage-detection model performance. In addition to combining data images from different sources with consistently relabeled damage instances, this study released road-damage image data from the road maintenance agency in Zhubei, Hsinchu County, Taiwan for research and other uses, increasing the limited amount of published image data sources and positively impacting future scholarly research into road damage detection. [ABSTRACT FROM AUTHOR]

License plate recognition is widely used in our daily life. Image binarization, which is a process to convert an image to white and black, is an important step of license plate recognition. Among the proposed binarization methods, Otsu method is the most famous and commonly used one in a license plate recognition system since it is the fastest and can reach a comparable recognition accuracy. The main disadvantage of Otsu method is that it is sensitive to luminance effect and noise, and this property is impractical since most vehicle images are captured in an open environment. In this paper, we propose a system to improve the performance of automatic license plates reorganization in the open environment in Taiwan. Our system uses a binarization method which is inspired by the symmetry principles. Experimental results showed that when our method has a similar time complexity to that of Otsu, our method can improve the recognition rate up to 1.30 times better than Otsu. [ABSTRACT FROM AUTHOR]

Fine aerosols with a diameter of less than 2.5 microns (PM2.5) have a significant negative impact on human health. However, their measurement devices or instruments are usually expensive and complicated operations are required, so a simple and effective way for measuring the PM2.5 concentration is needed. To relieve this problem, this paper attempts to provide an easy alternative approach to PM2.5 concentration estimation. The proposed approach is based on image processing schemes and a simple linear regression model. It uses images with a high and low PM2.5 concentration to obtain the difference between these images. The difference is applied to find the region with the greatest impact. The approach is described in two stages. First, a series of image processing schemes are employed to automatically select the region of interest (RoI) for PM2.5 concentration estimation. Through the selected RoI, a single feature is obtained. Second, by employing the single feature, a simple linear regression model is used and applied to PM2.5 concentration estimation. The proposed approach is verified by the real-world open data released by Taiwan's government. The proposed scheme is not expected to replace component analysis using physical or chemical techniques. We have tried to provide a cheaper and easier way to conduct PM2.5 estimation with an acceptable performance more efficiently. To achieve this, further work will be conducted and is summarized at the end of this paper. [ABSTRACT FROM AUTHOR]