A5028749038- Infrared Target Detection Methodologies
- Advanced Measurement and Detection Methods
- Ultrasonics and Acoustic Wave Propagation
- Advanced Neural Network Applications
- Non-Destructive Testing Techniques
- Geophysical Methods and Applications
- Optical Systems and Laser Technology
- 3D Shape Modeling and Analysis
- Remote Sensing and Land Use
- Video Surveillance and Tracking Methods
- Advanced Image Processing Techniques
- Advanced Image Fusion Techniques
- Robotics and Sensor-Based Localization
- Digital Media Forensic Detection
- 3D Surveying and Cultural Heritage
- Remote Sensing and LiDAR Applications
- Image and Object Detection Techniques
- Advanced Image and Video Retrieval Techniques
- Robot Manipulation and Learning
- Advanced Semiconductor Detectors and Materials
- Remote-Sensing Image Classification
- Visual Attention and Saliency Detection
- Image and Signal Denoising Methods
- Thermography and Photoacoustic Techniques
National University of Defense Technology
2023-2025
University of Strathclyde
2010-2012
National Chung Cheng University
2009
Network quantization is leveraged to reduce the model size, memory footprint, and computational cost of deep neural networks. It achieved by representing float weights activations with lower bit counterparts, which essential for deployment on resource-limited devices. However, due extremely small size infrared targets in feature map, low-bit could lead huge information loss thus causes severe segmentation performance degradation. To achieve while maintaining performance, we first study...
Moving object detection in satellite videos (SVMOD) is a challenging task due to the extremely dim and small target characteristics. Current learning-based methods extract spatio-temporal information from multi-frame dense representation with labor-intensive manual labels tackle SVMOD, which needs high annotation costs contains tremendous computational redundancy severe imbalance between foreground background regions. In this paper, we propose highly efficient unsupervised framework for...
Infrared small target (IRST) detection aims at separating targets from cluttered background. Although many deep learning-based single-frame IRST (SIRST) methods have achieved promising performance, they cannot deal with extremely dim while suppressing the clutters since are spatially indistinctive. Multiframe (MIRST) can well handle this problem by fusing temporal information of moving targets. However, extraction motion is challenging general convolution insensitive to direction. In...
The purpose of non-uniformity and blind pixel correction is to provide a more reliable foundation for subsequent image processing target detection. Existing methods generally struggle balance the contradiction between over-smoothing residual noise. Particularly, can easily filter out texture details dim small targets. Based on multi-frame response model infrared focal plane array detector, we propose two-stage 3-D fully convolutional network factor estimation, integrated with an suppression...
Visible-thermal small object detection (RGBT SOD) is a significant yet challenging task with wide range of applications, including video surveillance, traffic monitoring, search and rescue. However, existing studies mainly focus on either visible or thermal modality, while RGBT SOD rarely explored. Although some datasets have been developed, the insufficient quantity, limited diversity, unitary application, misaligned images large target size cannot provide an impartial benchmark to evaluate...
In ultrasonic NDT of heterogeneous materials the internal microstructure material produces backscattered noise that can make detection true defects difficult. The is caused by stationary scatterers cause constructive and destructive interference to propagating wavefront. Morevoer, in situations where are significantly larger than these random scatterers, limited due presence additive interfering scatterers. This causes speckle ultrasound imaging, thereby limiting detectability, making images...
During the last decade, industrial robots have been applied to various field of science from computer-assisted surgery space exploration. In recent years, autonomous service are increasingly finding their way into our daily lives. this paper, we present design robot, and robot arm path planning combined with several sensors: motor encoder, CCD camera laser range finder. Using information CCD-cameras that provide robot's environment which serves as an input for a robust object recognition...
Conventional ultrasonic C-scan imaging in composite materials typically employs peak amplitude extraction where the obtained image quality is extremely dependent on gate setup acquisition hardware. Additionally structural noise created by scattering phenomenon often high enough to bury meaningful reflection echoes, due non-homogeneous nature of such materials. This paper investigates use Bayesian inference as a method construct images material. The approach was implemented using experimental...
In this paper, a point cloud instance segmentation method is proposed to achieve infrared moving small multi-targets trajectory detection. Different from previous detection methods that used manually designed features frame-by-frame correlation, paper first models 100 frames of candidate images in 2D space along the time dimension as 3D space. way, we can transform problem correlation points segmenting instances with specific linear topological structures To end, propose simple but effective...