- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Image blind deblurring using an adaptive patch prior
摘要： Image blind deblurring uses an estimated blur kernel to obtain an optimal restored original image with sharp features from a degraded image with blur and noise artifacts. This method, however, functions on the premise that the kernel is estimated accurately. In this work, we propose an adaptive patch prior for improving the accuracy of kernel estimation. Our proposed prior is based on local patch statistics and can rebuild low-level features, such as edges, corners, and junctions, to guide edge and texture sharpening for blur estimation. Our prior is a nonparametric model, and its adaptive computation relies on internal patch information. Moreover, heuristic filters and external image knowledge are not used in our prior. Our method for the reconstruction of salient step edges in a blurry patch can reduce noise and over-sharpening artifacts. Experiments on two popular datasets and natural images demonstrate that the kernel estimation performance of our method is superior to that of other state-of-the-art methods.
关键词： low-level features,internal patch information,adaptive patch prior,blind deblurring,kernel estimation
更新于2025-09-23 15:22:29
-
Some Applications of ANN to Solar Radiation Estimation and Forecasting for Energy Applications
摘要： In solar energy, the knowledge of solar radiation is very important for the integration of energy systems in building or electrical networks. Global horizontal irradiation (GHI) data are rarely measured over the world, thus an artificial neural network (ANN) model was built to calculate this data from more available ones. For the estimation of 5-min GHI, the normalized root mean square error (nRMSE) of the 6-inputs model is 19.35%. As solar collectors are often tilted, a second ANN model was developed to transform GHI into global tilted irradiation (GTI), a difficult task due to the anisotropy of scattering phenomena in the atmosphere. The GTI calculation from GHI was realized with an nRMSE around 8% for the optimal configuration. These two models estimate solar data at time, t, from other data measured at the same time, t. For an optimal management of energy, the development of forecasting tools is crucial because it allows anticipation of the production/consumption balance; thus, ANN models were developed to forecast hourly direct normal (DNI) and GHI irradiations for a time horizon from one hour (h+1) to six hours (h+6). The forecasting of hourly solar irradiation from h+1 to h+6 using ANN was realized with an nRMSE from 22.57% for h+1 to 34.85% for h+6 for GHI and from 38.23% for h+1 to 61.88% for h+6 for DNI.
关键词： solar irradiation,estimation,meteorological data,short time step,forecasting
更新于2025-09-23 15:22:29
-
[IEEE 2018 IEEE International Conference on Consumer Electronics - Asia (ICCE-Asia) - Jeju (2018.6.24-2018.6.26)] 2018 IEEE International Conference on Consumer Electronics - Asia (ICCE-Asia) - Video Stabilization Using Feature-Based Classification
摘要： This paper proposed a video stabilization algorithm to cluster features before smoothing. Our method calculates the direction of movement from the features then estimate by directional statistics and K-mean to find out the global motion and motion of moving object. The motion will be smoothed by low pass Alpha-trimmed filter. The experimental show the effectiveness of our proposed method.
关键词： motion estimation,Video stabilization,clustering
更新于2025-09-23 15:22:29
-
An Improved DBSCAN Method for LiDAR Data Segmentation with Automatic Eps Estimation
摘要： Point cloud data segmentation, ?ltering, classi?cation, and feature extraction are the main focus of point cloud data processing. DBSCAN (density-based spatial clustering of applications with noise) is capable of detecting arbitrary shapes of clusters in spaces of any dimension, and this method is very suitable for LiDAR (Light Detection and Ranging) data segmentation. The DBSCAN method needs at least two parameters: The minimum number of points minPts, and the searching radius ε. However, the parameter ε is often harder to determine, which hinders the application of the DBSCAN method in point cloud segmentation. Therefore, a segmentation algorithm based on DBSCAN is proposed with a novel automatic parameter ε estimation method—Estimation Method based on the average of k nearest neighbors’ maximum distance—with which parameter ε can be calculated on the intrinsic properties of the point cloud data. The method is based on the ?tting curve of k and the mean maximum distance. The method was evaluated on different types of point cloud data: Airborne, and mobile point cloud data with and without color information. The results show that the accuracy values using ε estimated by the proposed method are 75%, 74%, and 71%, which are higher than those using parameters that are smaller or greater than the estimated one. The results demonstrate that the proposed algorithm can segment different types of LiDAR point clouds with higher accuracy in a robust manner. The algorithm can be applied to airborne and mobile LiDAR point cloud data processing systems, which can reduce manual work and improve the automation of data processing.
关键词： parameter estimation,segmentation,DBSCAN,LiDAR
更新于2025-09-23 15:22:29
-
[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) - Learning Illuminant Estimation from Object Recognition
摘要： In this paper we present a deep learning method to estimate the illuminant of an image. Our model is not trained with illuminant annotations, but with the objective of improving performance on an auxiliary task such as object recognition. To the best of our knowledge, this is the first example of a deep learning architecture for illuminant estimation that is trained without ground truth illuminants. We evaluate our solution on standard datasets for color constancy, and compare it with state of the art methods. Our proposal is shown to outperform most deep learning methods in a cross-dataset evaluation setup, and to present competitive results in a comparison with parametric solutions.
关键词： Illuminant estimation,deep learning,convolutional neural networks,computational color constancy,semi-supervised learning
更新于2025-09-23 15:22:29
-
Paddy acreage mapping and yield prediction using sentinel-based optical and SAR data in Sahibganj district, Jharkhand (India)
摘要： Rice is an important staple food for the billions of world population. Mapping the spatial distribution of paddy and predicting yields are crucial for food security measures. Over the last three decades, remote sensing techniques have been widely used for monitoring and management of agricultural systems. This study has employed Sentinel-based both optical (Sentinel-2B) and SAR (Sentinel-1A) sensors data for paddy acreage mapping in Sahibganj district, Jharkhand during the monsoon season in 2017. A robust machine learning Random Forest (RF) classification technique was deployed for the paddy acreage mapping. A simple linear regression yield model was developed for predicting yields. The key findings showed that the paddy acreage was about 68.3–77.8 thousand hectares based on Sentinel-1A and 2B satellite data, respectively. Accordingly, the paddy production of the district was estimated as 108–126 thousand tonnes. The paddy yield was predicted as 1.60 tonnes/hectare. The spatial distribution of paddy based on RF classifier and the accuracy assessment of LULC maps revealed that SAR-based classified paddy map was more consistent than the optical data. Nevertheless, this comprehensive study concluded that the SAR data could be more pronounced in acreage mapping and yield estimation for providing timely information to decision makers.
关键词： Yield estimation,SAR data,Acreage mapping,Random Forest classifier
更新于2025-09-23 15:22:29
-
Automatic robot path integration using three-dimensional vision and offline programming
摘要： In manufacturing industries, offline programming (OLP) platforms provide an independent methodology for robot integration using 3D model simulation away from the actual robot cell and production process, reducing integration time and costs. However, traditional OLP platforms still require prior knowledge of the workpiece position in a predefined environment, which requires complex human operations and specific-purpose designs, highly reducing the autonomy of the systems. The presented approach proposes to overcome these problems by defining a novel automated offline programming system (AOLP), which integrates a flexible and intuitive OLP platform with a state-of-the-art autonomous object pose estimation method, to achieve an environment and model independent platform for automatic robotic manufacturing. The autonomous recognition capabilities of the three-dimensional vision system provide the relative position of the workpiece model in the OLP platform, with robustness against clutter, illumination, and object material. After that, the user-friendly OLP platform allows an efficient and automatic path generation, simulation, robot code generation, and robot execution. The proposed system precision and robustness are analyzed and validated in a real-world environment on four different sets of experiment. Finally, the proposed system's features are discussed and compared with other available solutions for practical industrial manufacturing, showing the advantages of the proposed approach. Overall, despite sensor resolution limitations, the proposed system shows a remarkable precision and promising direction towards highly efficient and productive manufacturing solutions.
关键词： Machine vision,Path generation,Industrial manipulator,Automated offline programming,3D object recognition,6D pose estimation
更新于2025-09-23 15:22:29
-
Fast 6D object pose refinement in depth images
摘要： Recovering 6D object pose has gained much focus, because of its application in robotic intelligent manipulation to name but a few. This paper presents an approach for 6D object pose refinement from noisy depth images obtained from a consumer depth sensor. Compared to the state of the art aimed at the same goal, the proposed method has high precision, high robustness to partial occlusions and noise, low computation cost and fast convergence. This is achieved by using an iterative scheme that only employs Random Forest to minimize a cost function of object pose which can quantify the misalignment between the ground truth and the estimated one. The random forest in our algorithm is learnt only using synthetic depth images rendered from 3D model of the object. Several experimental results show the superior performance of the proposed approach compared to ICP-based algorithm and optimization-based algorithm, which are generally used for 6D pose refinement in depth images. Moreover, the iterative process of our algorithm can be much faster than the state of the art by only using one CPU core.
关键词： Object pose refinement,Random forest,Depth images,6D pose estimation,Fast convergence
更新于2025-09-23 15:22:29
-
Heart rate estimation from photoplethysmography signal for wearable health monitoring devices
摘要： Wearable wrist type health monitoring devices use photoplethysmography (PPG) signal to estimate heart rate (HR). The HR estimation from these devices becomes difficult due to the existence of strong motion artifacts (MA) in PPG signal thereby leading to inaccurate HR estimation. The objective is to develop a novel de-noising algorithm that reduces the MA present in PPG signal, resulting in an accurate HR estimation. A novel de-noising technique using the hierarchical structure of cascade and parallel combinations of two different pairs of adaptive filters which reduces MA from the PPG signal and improves HR estimation is proposed. The first pair combines normalized least mean squares (NLMS) and recursive least squares (RLS) adaptive filters and the second pair combines recursive least squares (RLS) and least mean squares (LMS) adaptive filters. The de-noised signals obtained from the first and second pairs are combined to form a single de-noised PPG signal by means of convex combination. The HR of the de-noised PPG signal is estimated in the frequency domain using a Fast Fourier transform (FFT). Performance of the proposed technique is evaluated using a dataset of 12 individuals performing running activity in Treadmill. It resulted in an average absolute error of 0.92 beats per minute (BPM), standard deviation of the absolute error of 1.17 beats per minute (BPM), average relative error of 0.72 and Pearson correlation coefficient of 0.9973.
关键词： Photoplethysmography,Convex combination,Heart rate estimation,Motion artifact,Wearable devices,Combination of adaptive filters
更新于2025-09-23 15:22:29
-
Hybrid TDOA/RSS based localization for visible light systems
摘要： In a visible light positioning (VLP) system, a receiver can estimate its location based on signals transmitted by light emitting diodes (LEDs). In this manuscript, we investigate a quasi-synchronous VLP system, in which the LED transmitters are synchronous among themselves but are not synchronized with the receiver. In quasi-synchronous VLP systems, position estimation can be performed by utilizing time difference of arrival (TDOA) information together with channel attenuation information, leading to a hybrid localization system. To specify accuracy limits for quasi-synchronous VLP systems, the Cramér–Rao lower bound (CRLB) on position estimation is derived in a generic three-dimensional scenario. Then, a direct positioning approach is adopted to obtain the maximum likelihood (ML) position estimator based directly on received signals from LED transmitters. In addition, a two-step position estimator is proposed, where TDOA and received signal strength (RSS) estimates are obtained in the first step and the position estimation is performed, based on the TDOA and RSS estimates, in the second step. The performance of the two-step positioning technique is shown to converge to that of direct positioning at high signal-to-noise ratios based on asymptotic properties of ML estimation. Finally, CRLBs and performance of the proposed positioning techniques are investigated through simulations.
关键词： Received signal strength (RSS),Estimation,Visible light,Localization,Time difference of arrival (TDOA)
更新于2025-09-23 15:22:29