- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[Advances in Intelligent Systems and Computing] Recent Findings in Intelligent Computing Techniques Volume 709 (Proceedings of the 5th ICACNI 2017, Volume 3) || Optimal Approach for Image Recognition Using Deep Convolutional Architecture
摘要： In the recent time, deep learning has achieved huge popularity due to its performance in various machine learning algorithms. Deep learning as hierarchical or structured learning attempts to model high-level abstractions in data by using a group of processing layers. The foundation of deep learning architectures is inspired by the understanding of information processing and neural responses in human brain. The architectures are created by stacking multiple linear or nonlinear operations. The article mainly focuses on the state-of-the-art deep learning models and various real-world application-speci?c training methods. Selecting optimal architecture for speci?c problem is a challenging task; at a closing stage of the article, we proposed optimal approach to deep convolutional architecture for the application of image recognition.
关键词： Deep neural networks,Image recognition,Image processing,Transfer learning,Convolutional neural networks,Deep learning
更新于2025-09-23 15:21:01
-
A data-driven model for weld bead monitoring during the laser welding assisted by magnetic field
摘要： In this research, a data-driven model is developed to monitor the seam during the laser beam welding under the influence of an external magnetic field (LBW-AMF). Firstly, a visible LBW-AMF system is built for tracking the laser melting pool and keyhole. Then, the features of the laser melting pool and keyhole are extracted with image processing techniques. The approach for an ensemble of different neural networks which includes radial basis function neural network, back-propagation neural network, and generalized regression neural network is proposed to establish the correlations of the characteristics of the laser melting pool and keyhole and the welding seam. Finally, LBW-AMF experimental results are obtained to validate the performance of the proposed data-driven model. Results illustrate that the developed model can provide a reliable result for monitoring the weld bead, which could give guidance for controlling the processing parameters in real time to improve the weld quality for practical LBW-AMF.
关键词： Image processing,Laser beam welding,Neural networks,Online monitoring,Data-driven model
更新于2025-09-23 15:21:01
-
Modulation Properties of an Extended Cavity Diode Laser and Dynamic Mode Splitting
摘要： Low-dose CT (LDCT) images tend to be degraded by excessive mottle noise and steak artifacts. In this paper, we proposed a novel fractional-order differentiation model that can be applied to LDCT image processing as a post-processing technique. The anisotropic diffusion model (proposed by Perona and Malik, i.e., PM model) has good performance in flat regions, total variation (TV) model works better in edge preservation, and fractional-order differentiation models can mitigate block effect while preserving fine details and more structure. The proposed model is based on the weighted combinations of the fractional-order PM model and the fractional-order TV model, which maintains the advantages of PM model, TV model, and fractional-order differentiation models. Moreover, the local intensity variance was added to both weighted coefficient and diffusion coefficient of the proposed model to properly preserve edges and details. A variety of simulated phantom data, including the Shepp–Logan head phantom, the pelvis phantom, and the actual thoracic phantom, were used for experimental validation. The results of numerical simulation and clinical data experiments demonstrate that the proposed approach has a better performance in both noise suppression and detail preservation, when compared with several other existing methods.
关键词： edge and detail preservation,fractional-order differentiation model,Low-dose CT,image processing
更新于2025-09-23 15:21:01
-
[IEEE 2019 IEEE 46th Photovoltaic Specialists Conference (PVSC) - Chicago, IL, USA (2019.6.16-2019.6.21)] 2019 IEEE 46th Photovoltaic Specialists Conference (PVSC) - High Open-Circuit Voltage in Wide-Bandgap Perovskite Photovoltaics with Passivation Layers Based on Large Cations
摘要： Segmentation of partially overlapping objects with a known shape is needed in an increasing amount of various machine vision applications. This paper presents a method for segmentation of clustered partially overlapping objects with a shape that can be approximated using an ellipse. The method utilizes silhouette images, which means that it requires only that the foreground (objects) and background can be distinguished from each other. The method starts with seedpoint extraction using bounded erosion and fast radial symmetry transform. Extracted seedpoints are then utilized to associate edge points to objects in order to create contour evidence. Finally, contours of the objects are estimated by ?tting ellipses to the contour evidence. The experiments on one synthetic and two different real data sets showed that the proposed method outperforms two current state-of-art approaches in overlapping objects segmentation.
关键词： image processing,overlapping objects,machine vision,convex objects,Segmentation
更新于2025-09-23 15:21:01
-
[IEEE 2019 2nd International Conference on Power Energy, Environment and Intelligent Control (PEEIC) - Greater Noida, India (2019.10.18-2019.10.19)] 2019 2nd International Conference on Power Energy, Environment and Intelligent Control (PEEIC) - Formulation of Efficiency of Inverters for Solar Photovoltaic Power Plants - Indian Case Study
摘要： Segmentation of partially overlapping objects with a known shape is needed in an increasing amount of various machine vision applications. This paper presents a method for segmentation of clustered partially overlapping objects with a shape that can be approximated using an ellipse. The method utilizes silhouette images, which means that it requires only that the foreground (objects) and background can be distinguished from each other. The method starts with seedpoint extraction using bounded erosion and fast radial symmetry transform. Extracted seedpoints are then utilized to associate edge points to objects in order to create contour evidence. Finally, contours of the objects are estimated by fitting ellipses to the contour evidence. The experiments on one synthetic and two different real data sets showed that the proposed method outperforms two current state-of-art approaches in overlapping objects segmentation.
关键词： image processing,overlapping objects,machine vision,convex objects,Segmentation
更新于2025-09-23 15:21:01
-
Solar hard X-ray imaging by means of compressed sensing and finite isotropic wavelet transform
摘要： Aims. Compressed sensing realized by means of regularized deconvolution and the finite isotropic wavelet transform is effective and reliable in hard X-ray solar imaging. Methods. The method uses the finite isotropic wavelet transform with the Meyer function as the mother wavelet. Furthermore, compressed sensing is realized by optimizing a sparsity-promoting regularized objective function by means of the fast iterative shrinkage-thresholding algorithm. Eventually, the regularization parameter is selected by means of the Miller criterion. Results. The method is applied against both synthetic data mimicking measurements made with the Spectrometer/Telescope Imaging X-rays (STIX) and experimental observations provided by the Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI). The performances of the method are qualitatively validated by comparing some morphological properties of the reconstructed sources with those of the corresponding synthetic configurations. Furthermore, the results concerning experimental data are compared with those obtained by applying other visibility-based reconstruction methods. Conclusions. The results show that when the new method is applied to synthetic STIX visibility sets, it provides reconstructions with a spatial accuracy comparable to the accuracy provided by the most popular method in hard X-ray solar imaging and with a higher spatial resolution. Furthermore, when it is applied to experimental RHESSI data, the reconstructions are characterized by reliable photometry and by a notable reduction of the ringing effects caused by the instrument point spread function.
关键词： techniques: image processing,Sun: X-rays, gamma rays,Sun: flares
更新于2025-09-23 15:21:01
-
Rapid detection of building cracks based on image processing technology with double square artificial marks
摘要： In order to measure the crack width of dangerous buildings quickly and accurately, this article presents a new crack width measurement method, which is based on image processing technology, using double square artificial markers to identify building cracks and calculate crack width. It makes two 10 mm 3 10 mm black square artificial marks and places them near the sides of the crack. Then it uses a camera to collect crack images and transfer photos to a computer. The crack image is subjected to image graying, binarization, denoising, image segmentation, and pixel calibration based on the image processing technique. Finally, the actual length value of the unit pixel is calculated. Then it can calculate the actual width of the crack according to the number of pixels included in the test crack. The calculation results show that the accuracy of the proposed method is 98.56% compared with the measured data. The calculation method can accurately and effectively detect the crack width of dangerous buildings and improve work efficiency. At the same time, it can avoid long hours of work in dangerous operating environments.
关键词： building cracks,digital image processing,safety,MATLAB,double square artificial marks,crack width
更新于2025-09-23 15:21:01
-
Tidal Features at 0.05 < <i>z</i> < 0.45 in the Hyper Suprime-Cam Subaru Strategic Program: Properties and Formation Channels
摘要： We present 1201 galaxies at 0.05 < z < 0.45 that host tidal features in the first ～200 deg2 of imaging from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). We select these galaxies from a sample of 21,208 galaxies with spectroscopic redshifts drawn from the Sloan Digital Sky Survey (SDSS) spectroscopic campaigns. Of these galaxies, we identify 214 shell systems and 987 stream systems. For 575 of these systems, we are additionally able to measure the (g?i) colors of the tidal features. We find evidence for star formation in a subset of the streams, with the exception of streams around massive ellipticals, and find that stream host galaxies span the full range of stellar masses in our sample. Galaxies that host shells are predominantly red and massive: we find that observable shells form more frequently around ellipticals than around disk galaxies of the same stellar mass. Although the majority of the shells in our sample are consistent with being formed by minor mergers, 15% ± 4.4% of shell host galaxies have (g?i) colors as red as their host galaxy, consistent with being formed by major mergers. These “red shell” galaxies are preferentially aligned with the major axis of the host galaxy, as previously predicted from simulations. We suggest that although the bulk of the observable shell population originates from fairly minor mergers, which preferentially form shells that are not aligned with the major axis of the galaxy, major mergers produce a significant number of observable shells.
关键词： techniques: image processing,galaxies: interactions,catalogs,galaxies: structure,galaxies: statistics
更新于2025-09-23 15:21:01
-
[Advances in Intelligent Systems and Computing] Intelligent Computing Volume 858 (Proceedings of the 2018 Computing Conference, Volume 1) || Using DSP-ASIP for Image Processing Applications
摘要： The rapid deployment of embedded image processing applications have forced a paradigm shift from complete hardware and software based implementations providing best performance and lowest cost, respectively towards a hybrid approach, namely, application specific instruction-set processor (ASIP). In this paper, we evaluate the applicability of CuSP, a softcore DSP-ASIP, for image processing applications. CuSP has a Crimson DSP processor core and hardware accelerators directly coupled with the core offering improved performance with flexibility. Results show that CuSP offers performance improvement over standard softprocessor MicroBlaze by up to a factor of 36 times. Crimson DSP core alone gives up to 5.3 times lower execution cycles than MicroBlaze.
关键词： Image processing,DSP-ASIP,2D convolution
更新于2025-09-23 15:21:01
-
[Advances in Intelligent Systems and Computing] Recent Findings in Intelligent Computing Techniques Volume 709 (Proceedings of the 5th ICACNI 2017, Volume 3) || Detection and Analysis of Oil Spill in Ocean for Reduced Complexity in Extraction Using Image Processing
摘要： Oil spills occurring in oceans are difficult to detect and require sophisticated measures to obtain and analyze the images. In this chapter, both color image using high-resolution cameras and Synthetic Aperture Radar (SAR) images are analyzed and certain useful results are obtained to reduce the complexity in extracting the oil spills. The recognition and examination of the oil spill images are done using image processing technique. Furthermore, if the oil spill is scattered as patches, the algorithm classifies the patches into smaller patches and larger ones by using k-means clustering. Hence, the patches depending on the size or intensity can be extracted on a simpler basis.
关键词： Image processing,Synthetic aperture radar (SAR) images,Machine learning,K-means clustering
更新于2025-09-23 15:21:01