scholarly journals Neural network texture segmentation of satellite images of woodlands using the U-net model

2021 ◽  
Author(s):  
A.E. Alyokhina ◽  
D.S. Rusin ◽  
E.V. Dmitriev ◽  
A.N. Safonova
Keyword(s):  
Neural Network ◽  
Forest Canopy ◽  
Structural Parameters ◽  
Spatial Scales ◽  
Texture Features ◽  
Texture Segmentation ◽  
Textural Analysis ◽  
Significant Regression ◽  
Contrast Normalization ◽  
The Fourier Transform

With the advent of space equipment that allows obtaining panchromatic images of ultra-high spatial resolution (< 1 m) there was a tendency to develop methods of thematic processing of aerospace images in the direction of joint use of textural and spectral features of the objects under study. In this paper, we consider the problem of classification of forest canopy structures based on textural analysis of multispectral and panchromatic images of Worldview-2. Traditionally, a statistical approach is used to solve this problem, based on the construction of distributions of the common occurrence of gray gradations and the calculation of statistical moments that have significant regression relationships with the structural parameters of stands. An alternative approach to solving the problem of extracting texture features is based on frequency analysis of images. To date, one of the most promising methods of this kind is based on wavelet scattering. In comparison with the traditionally applied approaches based on the Fourier transform, in addition to the characteristic signal frequencies, the wavelet analysis allows us to identify characteristic spatial scales, which is fundamentally important for the textural analysis of spatially inhomogeneous images. This paper uses a more general approach to solving the problem of texture segmentation using the convolutional neural network U-net. This architecture is a sequence of convolution-pooling layers. At the first stage, the sampling of the original image is lowered and the content is captured. At the second stage, the exact localization of the recognized classes is carried out, while the discretization is increased to the original one. The RMSProp optimizer was used to train the network. At the preprocessing stage, the contrast of fragments is increased using the global contrast normalization algorithm. Numerical experiments using expert information have shown that the proposed method allows segmenting the structural classes of the forest canopy with high accuracy.

scholarly journals Statistical texture analysis of forest areas from very high spatial resolution satellite images

2021 ◽  
Author(s):  
E.V. Dmitriev ◽  
T.V. Kondranin ◽  
P.G. Melnik ◽  
S.A. Donskoy
Keyword(s):  
Remote Sensing ◽  
Texture Analysis ◽  
Spatial Resolution ◽  
Forest Canopy ◽  
Structural Parameters ◽  
Texture Features ◽  
Mutual Arrangement ◽  
Environmental Objects ◽  
Remote Sensing Methods ◽  
Very High

Aerospace images with a spatial resolution of less than 1 m are actively used by regional services to obtain and update information about various environmental objects. Considerable efforts are being devoted to the development of remote sensing methods for forest areas. The structure of the forest canopy depends on various parameters, most of which are determined by ground-based methods during forest management works. Remote sensing methods for assessing the structural parameters of forest stands are based on texture analysis of panchromatic and multispectral images. A statistical approach is often used to extract texture features. The basis of this approach is the description of the distributions characterizing the mutual arrangement of image pixels in grayscale. This paper compares the effectiveness of matrix based statistical methods for extracting textural features for solving the problem of classifying various natural and manmade objects, as well as structures of the forest canopy. We consider statistics of various orders based on estimates of the distributions of gray levels, as well as the mutual occurrence, frequency, difference and structuring of gray levels. The results of assessing the informativeness of statistical textural characteristics in determining various structures of the forest canopy are presented. Dependences of the classification results on the choice of distribution parameters are determined. For the quantitative validation of the results obtained, data from ground surveys and expert visual classification of very high resolution WorldView-2 images of the territories of Savvatyevkoe and Bronnitskoe forestries are used.

Longitudinal vibration analysis and suppression of electric wheel system driven by in-wheel motor considering unbalanced magnetic pull

2018 ◽  
Vol 233 (11) ◽  
pp. 2729-2745 ◽  
Author(s):  
Shuguang Zuo ◽  
Duoqiang Li ◽  
Yu Mao ◽  
Wenzhe Deng
Keyword(s):  
Neural Network ◽  
Finite Element ◽  
Dynamic Model ◽  
Electric Vehicles ◽  
Bp Neural Network ◽  
Longitudinal Vibration ◽  
Structural Parameters ◽  
Torque Ripple ◽  
Longitudinal Dynamics ◽  
Unbalanced Magnetic Pull

With the blowout of electric vehicles recently, the key parts of the electric vehicles driven by in-wheel motors named the electric wheel system become the core of development research. The torque ripple of the in-wheel motor mainly results in the longitudinal dynamics of the electric wheel system. The excitation sources are first analyzed through the finite element method, including the torque ripple induced by the in-wheel motor and the unbalanced magnetic pull produced by the relative motion between the stator and rotor. The accuracy of the finite element model is verified by the back electromotive force test of the in-wheel motor. Second, the longitudinal-torsional coupled dynamic model is established. The proposed model can take into account the unbalanced magnetic pull. Based on the model, the modal characteristics and the longitudinal dynamics of the electric wheel system are analyzed. The coupled dynamic model is verified by the vibration test of the electric wheel system. Two indexes, namely, the root mean square of longitudinal vibration of the stator and the signal-to-noise ratio of the tire slip rate, are proposed to evaluate the electric wheel longitudinal performance. The influence of unbalanced magnetic pull on the evaluation indexes of the longitudinal dynamics is analyzed. Finally, the influence of motor’s structural parameters on the average torque, torque ripple, and equivalent electromagnetic stiffness are analyzed through the orthogonal test. A surrogate model between the structural parameters of the in-wheel motor and the average torque, torque ripple, and equivalent electromagnetic stiffness is established based on the Bp neural network. The torque ripple and the equivalent electromagnetic stiffness are then reduced through optimizing the structural parameters of the in-wheel motor. It turns out that the proposed Bp neural network–based method is effective to suppress the longitudinal vibration of the electric wheel system.

scholarly journals Integrated assessment of the state of sewage mine waters based on gas-discharge radiation method

2020 ◽  
Vol 201 ◽  
pp. 01032
Author(s):  
Natalija Glukhova ◽  
Viktor Khilov ◽  
Yuliia Kharlamova ◽  
Mariia Isakova
Keyword(s):  
Fourier Transform ◽  
Integrated Assessment ◽  
Experimental Studies ◽  
Texture Features ◽  
Biological Properties ◽  
Gas Discharge ◽  
Mine Waters ◽  
Physical Chemical ◽  
Spatial Frequencies ◽  
The Fourier Transform

The existing environmental problems resulting from the use and subsequent discharge of water by mining and metallurgical enterprises are analyzed. The relevance of developing methods for an integrated assessment of water properties is presented, which will allow the study of not only physical-chemical, but also the biological parameters of water. Experimental studies of water samples are conducted based on the method of gas-discharge radiation. The existing methods for extracting informative features of gas-discharge radiation images of liquid-phase objects are analyzed. Histograms of gas-discharge radiation images of various types of water were constructed and studied. The expediency of dividing images into separate fragments, characterized by common texture features, is shown. The analysis of texture features is carried out based on the use of the Fourier transform. Using the Fourier transform for the corresponding luminance vectors, the spectra have been obtained of the spatial distribution of the frequencies of brightness changes, which are used as texture features to identify the specific characteristics of the gas-discharge tracks formation. It has been determined that the developed method of water analysis allows assessing its biological properties, which is based on the values of the spatial frequency range. The advantage of the proposed method over the existing ones is the possibility of quantitative assessment in the form of numerical values of the covered range of spatial frequencies. The proposed method of studying the biological properties of water can be used as part of modern environmental monitoring systems.

scholarly journals Spatial Variation in Canopy Structure across Forest Landscapes

2018 ◽  
Author(s):  
Brady S. Hardiman ◽  
Elizabeth A. LaRue ◽  
Jeff W. Atkins ◽  
Robert T. Fahey ◽  
Franklin W. Wagner ◽  
...  
Keyword(s):  
Temperate Forest ◽  
Forest Canopy ◽  
Spatial Scales ◽  
Canopy Structure ◽  
Ecosystem Functions ◽  
Lidar System ◽  
Stable Point ◽  
Eastern Usa ◽  
Forested Landscapes ◽  
Patch Scale

Forest canopy structure (CS) controls many ecosystem functions and is highly variable across landscapes, but the magnitude and scale of this variation is not well understood. We used a portable canopy lidar system to characterize variation in five categories of CS along N = 3 transects (140&ndash;800 m long) at each of six forested landscapes within the eastern USA. The cumulative coefficient of variation was calculated for subsegments of each transect to determine the point of stability for individual CS metrics. We then quantified the scale at which CS is autocorrelated using Moran&rsquo;s I in an Incremental Autocorrelation analysis. All CS metrics reached stable values within 300 m but varied substantially within and among forested landscapes. A stable point of 300 m for CS metrics corresponds with the spatial extent that many ecosystem functions are measured and modeled. Additionally, CS metrics were spatially autocorrelated at 40 to 88 m, suggesting that patch scale disturbance or environmental factors drive these patterns. Our study shows CS is heterogeneous across temperate forest landscapes at the scale of 10&rsquo;s of meters, requiring a resolution of this size for upscaling CS with remote sensing to large spatial scales.

scholarly journals Fruit Disease Detection Using Convolution Neural Network Approach

2018 ◽  
Vol 7 (2) ◽  
pp. 62-65
Author(s):  
Shivani . ◽  
Sharanjit Singh
Keyword(s):  
Neural Network ◽  
Early Stage ◽  
Texture Features ◽  
Convolution Neural Network ◽  
Image Feature ◽  
Disease Detection ◽  
Neural Network Approach ◽  
Disease Information ◽  
Fruit Disease ◽  
Farming Industry

Fruit disease detection is critical at early stage since it will affect the farming industry. Farming industry is critical for the growth of the economic conditions of India. To this end, proposed system uses universal filter for the enhancement of image captured from source. This filter eliminates the noise if any from the image. This filter is not only tackle’s salt and pepper noise but also Gaussian noise from the image. Feature extraction operation is applied to extract colour and texture features. Segmented image so obtained is applied with Convolution neural network and k mean clustering for classification. CNN layers are applied to obtain optimised result in terms of classification accuracy. Clustering operation increases the speed with which classification operation is performed. The clusters contain the information about the disease information. Since clusters are formed so entire feature set is not required to be searched. Labelling information is compared against the appropriate clusters only. Results are improved by significant margin proving worth of the study.

A Deep Neural Network Model for Learning Runtime Frequency Response Function Using Sensor Measurements

2021 ◽  
Author(s):  
Yongzhi Qu ◽  
Gregory W. Vogl ◽  
Zechao Wang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Fourier Transform ◽  
Time Domain ◽  
Frequency Response Function ◽  
Frequency Component ◽  
Operating Conditions ◽  
Input Output ◽  
The Fourier Transform ◽  
The Time Domain

Abstract The frequency response function (FRF), defined as the ratio between the Fourier transform of the time-domain output and the Fourier transform of the time-domain input, is a common tool to analyze the relationships between inputs and outputs of a mechanical system. Learning the FRF for mechanical systems can facilitate system identification, condition-based health monitoring, and improve performance metrics, by providing an input-output model that describes the system dynamics. Existing FRF identification assumes there is a one-to-one mapping between each input frequency component and output frequency component. However, during dynamic operations, the FRF can present complex dependencies with frequency cross-correlations due to modulation effects, nonlinearities, and mechanical noise. Furthermore, existing FRFs assume linearity between input-output spectrums with varying mechanical loads, while in practice FRFs can depend on the operating conditions and show high nonlinearities. Outputs of existing neural networks are typically low-dimensional labels rather than real-time high-dimensional measurements. This paper proposes a vector regression method based on deep neural networks for the learning of runtime FRFs from measurement data under different operating conditions. More specifically, a neural network based on an encoder-decoder with a symmetric compression structure is proposed. The deep encoder-decoder network features simultaneous learning of the regression relationship between input and output embeddings, as well as a discriminative model for output spectrum classification under different operating conditions. The learning model is validated using experimental data from a high-pressure hydraulic test rig. The results show that the proposed model can learn the FRF between sensor measurements under different operating conditions with high accuracy and denoising capability. The learned FRF model provides an estimation for sensor measurements when a physical sensor is not feasible and can be used for operating condition recognition.

Sign in / Sign up

Export Citation Format

Share Document