- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 3rd Russian-Pacific Conference on Computer Technology and Applications (RPC) - Vladivostok (2018.8.18-2018.8.25)] 2018 3rd Russian-Pacific Conference on Computer Technology and Applications (RPC) - Image Analysis Based on Salient Points of Wavelet Transform
摘要： In the article the task of image analysis based on salient points of wavelet transform is considered. Salient points retrieval based on energy estimation of wavelet transform is described. Salient points description based on local binary patterns are proposed.
关键词： wavelet transform,segmentation,salient points,images matching
更新于2025-09-23 15:21:01
-
KNN-Based Representation of Superpixels for Hyperspectral Image Classification
摘要： Superpixel segmentation has been demonstrated to be a powerful tool in hyperspectral image (HSI) classification. Each superpixel region can be regarded as a homogeneous region, which is composed of a series of spatial neighboring pixels. However, a superpixel region may contain the pixels from different classes. To further explore the optimal representations of superpixels, a new framework based on two k selection rules is proposed to find the most representative training and test samples. The proposed method consists of the following four steps: first, a superpixel segmentation algorithm is performed on the HSI to cluster the pixels with similar spectral features into the same superpixel. Then, a domain transform recursive filtering is used to extract the spectral–spatial features of the HSI. Next, the k nearest neighbor (KNN) method is utilized to select k1 representative training samples and k2 test pixels for each superpixel, which can effectively overcome the within-class variations and between-class interference, respectively. Finally, the class label of superpixels can be determined by measuring the averaged distances among the selected training and test samples. Experiments conducted on four real hyperspectral datasets show that the proposed method provides competitive classification performances with respect to several recently proposed spectral–spatial classification methods.
关键词： superpixel segmentation,hyperspectral image classification,k nearest neighbor (KNN),Domain transform recursive filtering (RF)
更新于2025-09-23 15:21:01
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Ransac-Based Segmentation for Building Roof Face Detection in Lidar Point Cloud
摘要： This work proposes a method for segmenting the roof planes of buildings in Light Detection and Ranging (LiDAR) data. First, a preprocessing of the point cloud is performed to separate the points belonging to each building. The RANdom SAmple Consensus (RANSAC) method is then used in each building region to identify sets of coplanar points belonging to the roof faces. Finally, planar segments representing the same roof face are connected to minimize the fragmentation that may occur in the previous step. This requires the use of techniques for analyzing the continuity of adjacent planar segments. Although several thresholds are required, they can be predetermined or adapted, thus avoiding their modification by an operator in each application of the method. The results show that the proposed method functions appropriately, rarely failing in regions affected by local structures such as trees and antennas. Consequently, average rates higher than 90% were obtained for completeness and correction.
关键词： RANSAC,roof segmentation,LiDAR
更新于2025-09-23 15:21:01
-
Variability and repeatability of quantitative uptake metrics in [ <sup>18</sup> F]FDG PET/CT imaging of non-small cell lung cancer: impact of segmentation method, uptake interval, and reconstruction protocol
摘要： OBJECTIVES: There is increased interest in various new quantitative uptake metrics beyond standardized uptake value (SUV) in oncology PET/CT studies. The purpose of this study is to investigate the variability and test-retest repeatability (TRT) of metabolically active tumor volume (MATV) measurements and several other new quantitative metrics in non-small cell lung cancer (NSCLC) using [18F]FDG PET/CT with different segmentation methods, user interactions, uptake intervals, and reconstruction protocols. METHODS: Ten advanced NSCLC patients received two whole-body [18F]FDG PET/CT scans at both 60 and 90 min post-injection. PET data were reconstructed with four different protocols. Eight segmentation methods were applied to delineate lesions with and without a tumor mask. MATV, maximum and mean SUV (SUVmax, SUVmean), total lesion glycolysis (TLG), and intralesional heterogeneity features were derived. Variability and repeatability were evaluated using a generalized estimating equations statistical model with Bonferroni correction for multiple comparisons. The statistical model, including interaction between uptake interval and reconstruction protocol, was applied individually to the data obtained from each segmentation method. RESULTS: Without masking, none of the segmentation methods could delineate all lesions correctly. MATV was affected by both uptake interval and reconstruction settings for most segmentation methods. Similar observations were obtained for the uptake metrics SUVmax, SUVmean, TLG, homogeneity, entropy, and zone percentage. No effect of uptake interval was observed on TRT metrics, while the reconstruction protocol affected the TRT of SUVmax. Overall, segmentation methods showing poor quantitative performance in one condition showed better performance in other (combined) conditions. For some metrics, a clear statistical interaction was found between the segmentation method and both uptake interval and reconstruction protocol. CONCLUSIONS: All segmentation results need to be reviewed critically. MATV and other quantitative uptake metrics, as well as their TRT, depend on segmentation method, uptake interval, and reconstruction protocol. To obtain quantitative reliable metrics, with good TRT performance, the optimal segmentation method depends on local imaging procedure, the PET/CT system and/or reconstruction protocol. Rigid harmonization of imaging procedure and PET/CT performance will be helpful in mitigating this variability.
关键词： non-small cell lung cancer,segmentation method,positron emission tomography imaging,repeatability,Variability
更新于2025-09-23 15:21:01
-
[ACM Press the 2018 International Conference - Hong Kong, Hong Kong (2018.02.24-2018.02.26)] Proceedings of the 2018 International Conference on Image and Graphics Processing - ICIGP 2018 - The Optimized Level Set Image Segmentation Based on Saliency Maps
摘要： In order to improve the practicability of the level set method and reduce the computational cost, a optimized level set active contour model that embeds the image local information is proposed in this paper. Firstly, the optimal saliency method of the image is selected by comparing three saliency methods, which is helped to generate the initial contour of the image. Secondly, a new variational level set model integrating edge information and regional local information is presented, and a local energy term is added to the energy function. Finally, image segmentation is implemented by new level set methods based on optimal saliency method. Experiments demonstrate the results of the new level set method based on optimal saliency method are higher than the Distance Regularized Level Set Evolution (DRLSE) model in terms of both efficiency and accuracy. Moreover, the segmentation time of the optimization algorithm only needs 1.94% of the former, and it has high segmentation accuracy.
关键词： the level set method,Image segmentation,Saliency map,Distance Regularized Level Set Evolution (DRLSE)
更新于2025-09-23 15:21:01
-
[IEEE 2018 IEEE 18th International Power Electronics and Motion Control Conference (PEMC) - Budapest, Hungary (2018.8.26-2018.8.30)] 2018 IEEE 18th International Power Electronics and Motion Control Conference (PEMC) - Novel Segmentation Algorithm for Maximum Power Point Tracking in PV Systems Under Partial Shading Conditions
摘要： The paper proposes new MPPT Segmentation method for partial shading conditions. The proposed method is based on principle of division of the power-voltage curve into uniform segments and location of global maximum power point inside the chosen segment. In order to validate the algorithm, it was tested under different shading condition and with different PV systems. The simulation results show that the Segmentation algorithm precisely identifies the highest MPP on the power-voltage curve. The Segmentation algorithm was compared with standard P&O and also with other MPPT algorithms for partial shading conditions. This comparison showed that Segmentation algorithm provides better results and insures higher output power of the PV system than standard P&O and also with other MPPT algorithms. Simulation results also show that Segmentation algorithm works faster than P&O.
关键词： Segmentation algorithm,power-voltage curve,partial shading conditions,PV systems,MPPT
更新于2025-09-23 15:21:01
-
[IEEE 2018 25th IEEE International Conference on Image Processing (ICIP) - Athens, Greece (2018.10.7-2018.10.10)] 2018 25th IEEE International Conference on Image Processing (ICIP) - Sampling Technique for Defining Segmentation Error Margins with Application to Structural Brain Mri
摘要： Image segmentation is often considered a deterministic process with a single ground truth. Nevertheless, in practice, and in particular, when medical imaging analysis is considered, the extraction of regions of interest (ROIs) is ill-posed and the concept of 'most probable' segmentation is model-dependent. In this paper, a measure for segmentation uncertainty in the form of segmentation error margins is introduced. This measure provides a goodness quantity and allows a 'fully informed' comparison between extracted boundaries of related ROIs as well as more meaningful statistical analysis. The tool we present is based on a novel technique for segmentation sampling in the Fourier domain and Markov Chain Monte Carlo (MCMC). The method was applied to cortical and sub-cortical structure segmentation in MRI. Since the accuracy of segmentation error margins cannot be validated, we use receiver operating characteristic (ROC) curves to support the proposed method. Precision and recall scores with respect to expert annotations suggest this method as a promising tool for a variety of medical imaging applications including user-interactive segmentation, patient follow-up, and cross-sectional analysis.
关键词： Fourier domain,Segmentation uncertainty margins,sampling,MRI,Markov Chain Monte Carlo
更新于2025-09-23 15:21:01
-
Vine Signal Extraction – an Application of Remote Sensing in Precision Viticulture
摘要： This paper presents a study of precision agriculture in the wine industry. While precision viticulture mostly aims to maximise yields by delivering the right inputs to appropriate places on a farm in the correct doses and at the right time, the objective of this study was rather to assess vine biomass differences. The solution proposed in this paper uses aerial imagery as the primary source of data for vine analysis. The first objective to be achieved by the solution is to automatically identify vineyards blocks, vine rows, and individual vines within rows. This is made possible through a series of enhancements and hierarchical segmentations of the aerial images. The second objective is to determine the correlation of image data with the biophysical data (yield and pruning mass) of each vine. A multispectral aerial image is used to compute vegetation indices, which serve as indicators of biophysical measures. The results of the automatic detection are compared against a test field, to verify both vine location and vegetation index correlation with relevant vine parameters. The advantage of this technique is that it functions in environments where active cover crop growth between vines is evident and where variable vine canopy conditions are present within a vineyard block.
关键词： precision viticulture,remote sensing,segmentation,GIS,Precision agriculture,classification
更新于2025-09-23 15:21:01
-
Automatic lung segmentation in low-dose chest CT scans using convolutional deep and wide network (CDWN)
摘要： Computed tomography (CT) imaging is the preferred imaging modality for diagnosing lung-related complaints. Automatic lung segmentation is the most common prerequisite to develop a computerized diagnosis system for analyzing chest CT images. In this paper, a convolutional deep and wide network (CDWN) is proposed to segment lung region from the chest CT scan for further medical diagnosis. Earlier lung segmentation techniques depend on handcrafted features, and their performance relies on the features considered for segmentation. The proposed model automatically segments the lung from complete CT scan in two laps: (1) learning the required ?lters to extract hierarchical feature representations at convolutional layers, (2) dense prediction with spatial features through learnable deconvolutional layers. The model has been trained and evaluated with low-dose chest CT scan images on LIDC-IDRI database. The proposed CDWN reaches the average Dice coef?cient of 0.95 and accuracy of 98% in segmenting the lung regions from 20 test images and maintains consistent results for all test images. The experimental results con?rm that the proposed approach achieves a superior performance compared to other state-of-the-art methods for lung segmentation.
关键词： Medical imaging,Image processing and analysis,Deep learning,Automatic lung segmentation,Convolutional neural network
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) - Understanding hole transport across amorphous Si passivation layers in Si heterojunction solar cells using Monte Carlo simulation
摘要： Deformable models and level set methods have been extensively investigated for computerized image segmentation. However, medical image segmentation is yet one of open challenges owing to diversified physiology, pathology, and imaging modalities. Existing level set methods suffer from some inherent drawbacks in face of noise, ambiguity, and inhomogeneity. It is also refractory to control level set segmentation that is dependent on image content and evolutional strategies. In this paper, a new level set formulation is proposed by using fuzzy region competition for selective image segmentation. It is able to detect and track the arbitrary combination of selected objects or image components. To the best of our knowledge, this new formulation should be one of the first proposals in a framework of region competition for selective segmentation. Experiments on both synthetic and real images validate its advantages in selective level set segmentation.
关键词： image segmentation,region competition,level set methods,Fuzzy control,selective segmentation
更新于2025-09-23 15:19:57