DAE-MLP Based Feature Extraction for Hyperspectral Image Classification of Saint Clair River

Airborne multispectral and hyperspectral imaging can be used to detect potentially diseased trees rapidly over a large area using unique spectral signatures. Ground inspection and management can be focused on these detected zones, rather than an entire grove, making it less labor-intensive and time-consuming. We propose a method to detect the areas of citrus groves infected with citrus greening disease [Huanglongbing (HLB)] using airborne hyperspectral and multispectral imaging. This would prevent further spread of infection with efficient management plans of infected areas. Two sets of hyperspectral images were acquired in 2007 and 2009, from different citrus groves in Florida. Multispectral images were acquired only in 2009. A comprehensive ground truthing based on ground measurements and visual check of the citrus trees was used for validating the results using 2007 images. In 2009, a more accurate polymerase chain reaction test for selected trees from ground truthing was carried out. With a handheld spectrometer, ground spectral measurements were obtained along with their degrees of infection. A hyperspectral imaging software (ENVI, ITT VIS) was used for the analysis. HLB infected areas were identified using image-derived spectral library, mixture tuned matched filtering (MTMF), spectral angle mapping (SAM), and linear spectral unmixing. The accuracy of the MTMF method was greater than the other methods. The accuracy of SAM using multispectral images (87%) was comparable to the results of the MTMF and also yielded higher accuracy when compared to SAM analysis on hyperspectral images. A possible inaccurate ground truthing for the grove in 2007 generated more false positives.

Spectral angle mapper (SAM) has been widely used as a spectral similarity measure for multispectral and hyperspectral image analysis. It has been shown to be equivalent to Euclidean distance when the spectral angle is relatively small. Most recently, a stochastic measure, called spectral information divergence (SID) has been introduced to model the spectrum of a hyperspectral image pixel as a probability distribution so that spectral variations can be captured more effectively in a stochastic manner. This paper develops a new hyperspectral spectral discriminant measure, which is a mixture of SID and SAM. More specifically, let x_i and x_j denote two hyperspectral image pixel vectors with their corresponding spectra specified by s_i and s_j. SAM is the spectral angle of x_i and x_j and is defined by [SAM(s_i,s_j)]. Similarly, SID measures the information divergence between x_i and x_j and is defined by [SID(s_i,s_j)]. The new measure, referred to as (SID,SAM)-mixed measure has two variations defined by SID(s_i,s_j)xtan(SAM(s_i,s_j)] and SID(s_i,s_j)xsin[SAM(s_i,s_j)] where tan [SAM(s_i,s_j)] and sin[SAM(s_i,s_j)] are the tangent and the sine of the angle between vectors x and y. The advantage of the developed (SID,SAM)-mixed measure combines both strengths of SID and SAM in spectral discriminability. In order to demonstrate its utility, a comparative study is conducted among the new measure, SID and SAM where the discriminatory power of the (SID,SAM)-mixed measure is significantly improved over SID and SAM.

PurposeMost of the redundant dual-arm robots are singular free, dexterous and collision free compared to other robotic arms. This paper aims to analyse the workspace of redundant arms to study the manipulability. Furthermore, multi-layer perceptron (MLP) algorithm is used to determine the various joint parameters of both the upper body redundant arms. Trajectory planning of robotic arms is carried out with the help of inverse solutions obtained from the MLP algorithm.Design/methodology/approachIn this paper, the kinematic equations are derived from screw theory approach and inverse kinematic solutions are determined using MLP algorithm. Levenberg–Marquardt (LM) and Bayesian regulation (BR) techniques are used as the backpropagation algorithms. The results from two backpropagation techniques are compared for determining the prediction accuracy. The inverse solutions obtained from the MLP algorithm are then used to optimize the cubic spline trajectories planned for avoiding collision between arms with the help of convex optimization technique. The dexterity of the redundant arms is analysed with the help of Cartesian workspace of arms.FindingsDexterity of redundant arms is analysed by studying the voids and singular spaces present inside the workspace of arms. MLP algorithms determine unique solutions with less computational effort using BR backpropagation. The inverse solutions obtained from MLP algorithm effectively optimize the cubic spline trajectory for the redundant dual arms using convex optimization technique.Originality/valueMost of the MLP algorithms used for determining the inverse solutions are used with LM backpropagation technique. In this paper, BR technique is used as the backpropagation technique. BR technique converges fast with less computational time than LM method. The inverse solutions of arm joints for traversing optimized cubic spline trajectory using convex optimization technique are computed from the MLP algorithm.

ABSTRACTWith the large number of spectral bands in hyperspectral images, the conventional classification methods commonly used for multispectral images are not effectively applicable. To overcome such difficulty, feature extraction methods could be used to reduce the dimension of hyperspectral images. In this study, the performance of the principal component analysis (PCA) as a widely used technique in feature extraction and the wavelet transform as a powerful decomposition tool on hyperspectral data is compared. In wavelet transform, a non-linear wavelet feature extraction was employed to select efficient features for more classification accuracy. Shortwave infrared bands of Hyperion imagery were selected as input data. The study area includes two well-known porphyry copper deposits, Darrehzar and Sarcheshmeh, located in the Iranian copper belt. Neural networks (NN), Support Vector Machine (SVM), and Spectral Angle Mapper (SAM) were used for multi-class classification based on hydrothermal alteration zones and then trained by mineral spectral features related to typical porphyry copper deposits. In the NN set-up used in this study, one hidden layer was used, with the number of neurons equal to the number of features in the input layer. Conjugate gradient backpropagation was employed as the network training function. Then, the efficiency of feature extraction methods was compared through their classification accuracies. According to the results, although the highest classification accuracy for the PCA method occurs in lower numbers of extracted features compared to wavelet transform, the wavelet transform outperforms the PCA, based on confusion matrix classification. Moreover, NN is stronger than SVM and SAM in discriminating favourable alteration zones associated with porphyry copper mineralization using hyperspectral images.

Head blight on wheat, caused by Fusarium spp., is a serious problem for both farmers and food production due to the concomitant production of highly toxic mycotoxins in infected cereals. For selective mycotoxin analyses, information about the on-field status of infestation would be helpful. Early symptom detection directly on ears, together with the corresponding geographic position, would be important for selective harvesting. Hence, the capabilities of various digital imaging methods to detect head blight disease on winter wheat were tested. Time series of images of healthy and artificially Fusarium-infected ears were recorded with a laboratory hyperspectral imaging system (wavelength range: 400 nm to 1,000 nm). Disease-specific spectral signatures were evaluated with an imaging software. Applying the ‘Spectral Angle Mapper’ method, healthy and infected ear tissue could be clearly classified. Simultaneously, chlorophyll fluorescence imaging of healthy and infected ears, and visual rating of the severity of disease was performed. Between six and eleven days after artificial inoculation, photosynthetic efficiency of infected compared to healthy ears decreased. The severity of disease highly correlated with photosynthetic efficiency. Above an infection limit of 5% severity of disease, chlorophyll fluorescence imaging reliably recognised infected ears. With this technique, differentiation of the severity of disease was successful in steps of 10%. Depending on the quality of chosen regions of interests, hyperspectral imaging readily detects head blight 7 d after inoculation up to a severity of disease of 50%. After beginning of ripening, healthy and diseased ears were hardly distinguishable with the evaluated methods.

In the realm of stock market prediction and classification, the use of machine learning algorithms has gained significant attention. In this study, we explore the application of Multilayer Perceptron (MLP) and Radial Basis Function (RBF) algorithms in predicting and classifying stock prices, specifically amidst economic policy uncertainty. Stock market fluctuations are greatly influenced by economic policies implemented by governments and central banks. These policies can create uncertainty and volatility, which in turn makes accurate predictions and classifications of stock prices more challenging. By leveraging MLP and RBF algorithms, we aim to develop models that can effectively navigate these uncertainties and provide valuable insights to investors and financial analysts. The MLP algorithm, based on artificial neural networks, is able to learn complex patterns and relationships within financial data. The RBF algorithm, on the other hand, utilizes radial basis functions to capture non-linear relationships and identify hidden patterns within the data. By combining these algorithms, we aim to enhance the accuracy of stock price prediction and classification models. The results showed that both MLB and RBF predicted stock prices well for a group of countries using an index reflecting the impact of news on economic policy and expectations, where the MLB algorithm proved its ability to predict chain data. Countries were also classified according to stock price data and uncertainty in economic policy, allowing us to determine the best country to invest in according to the data. The uncertainty surrounding economic policy is what makes stock price forecasting so crucial. Investors must consider the degree of economic policy uncertainty and how it affects asset prices when deciding how to allocate their assets.

This paper explains how to use the SAM method for satellite image categorization and how to implement it. Hyperspectral Images (HI) provide diverse pixel spectrums that retrieve rich surface information for visualization and differentiation of spectrally similar (but distinct) items. By the HI sensor, HI with multiple bands of spectrum are captured on the Remote Sensing Satellite (RSS). For the deployment of the Spectral Angle Mapper (SAM) on Hyperspectral Images, classifications are performed. In the different surfaces the fake color composite of the image is further obtained for better observation. For SAM implementation, HI of various bands are piled one after another in the form a three-dimensional Cube of images. A supervised classification technique SAM that recognizes the various classes within an image, allowing the spectral angle to be calculated. The spectral angle can be determined using the vector created for each pixel and, as a result, the reference vector created for each of the reference classes chosen. The outcomes are acquired by combining several 2-D data to 3-D data with single compact cube and reading and observing them. Because of the reference vector is used for SAM classification, the angle between the reference vector and the pixel vector is calculated to match the threshold angle value. The color coding is then used to distinguish between the several classes that the SAM algorithm recognizes. As a result, SAM is used to analyze hyperspectral photos in order to thematic information extract such as land water bodies, cover, and clouds.

Target detection is one of research focuses for precision chemical application. This study developed a method to identify seedling cabbages and weeds using hyperspectral imaging. In processing the image data with ENVI software, after dimension reduction, noise reduction, de-correlation for high-dimensional data, and selection of the region of interest, the SAM (Spectral Angle Mapping) model was built for automatic identification of cabbages and weeds. With the HSI (Hyper Spectral Imaging) Analyzer, the training pixels were used to calculate the average spectrum as the standard spectrum. The parameters of the SAM model, which had the best classification results with 3-point smoothing, zero-order derivative, and 6-degrees spectral angle, was determined to achieve the accurate identification of the background, weeds, and cabbages. In comparison, the SAM model can completely separate the plants from the soil background but not perfect for weeds to be separated from the cabbages. In conclusion, the SAM classification model with the HSI analyzer could completely distinguish weeds from background and cabbages. Keywords: hyperspectral imaging, weed identification, cabbage, seedlings DOI: 10.3965/j.ijabe.20150805.1492 Citation: Deng W, Huang Y B, Zhao C J, Wang X. Identification of seedling cabbages and weeds using hyperspectral imaging. Int J Agric & Biol Eng, 2015; 8(5): 65－72.

Hyperspectral imaging is a modality which combines spatial resolution and spectroscopy in one technique. Analysis of hyperspectral data from biological samples is a demanding task due to the large amount of data, and due to the complex optical properties of biological tissue. In this study it was investigated if depth information could be revealed from hyperspectral images using a combination of image analysis and analytic simulations of skin reflectance. It was also investigated if hyperspectral imaging could be utilized to monitor changes in the distribution of hemoglobin species after smoking. Hyperspectral data in the wavelength range 400-1000nm were collected from the forearm of 15 non-smokers and 5 smokers. The hyperspectral images were analyzed with respect to the distribution of hemoglobin species and vascular structures. Changes in the vascular system due to smoking were also evaluated. Principal component analysis (PCA), Spectral angle mapping (SAM), and Mixture tuned matched filtering (MTMF) were used to enhance vascular structures. Emphasis was put on identifying apparent and true absorption spectra for the present chromophores by combining image analysis and an analytical photon transport model. The results show that the depth resolution of hyperspectral images can be better understood using analytical simulations. Modulation of the chromophore spectra by the optical properties of overlying tissue was found to be an important mechanism causing the depth resolution in hyperspectral images. It was also found that hyperspectral imaging and image analysis can be successfully applied to quantify skin changes following smoking.

Imaging spectroscopy in the remote sensing is an ever emerging platform that has offered the hyperspectral imaging (HSI) which delivers the Earth’s object information in hundreds of bands. HSI integrates conventional imaging with spectroscopy to get rich spectral and spatial features of the object. However, the challenges associated with HSI are its huge dimensionality and data redundancy that requests huge space, complex computations, and lengthier processing time. Therefore, this study aims to find the optimal bands to characterize the roof surfaces using supervised classifiers. To deal with high dimensionality of hyperspectral data, this study assesses the band selection method over data transformation methods. This study provides the comparison between data reduction methods and used classifiers. The height information from LiDAR was used to characterize urban roofs above the height of 2.5 meters. The optimal bands were investigated using supervised classifiers such as artificial neural network (ANN), support vector machine (SVM), and spectral angle mapper (SAM) by comparing accuracies. The classification result shows that ANN and SVM classifiers outperform whereas SAM performed poorly in roof characterization. The band selection method worked efficiently than the transformation methods. The classification algorithm successfully identifies the optimum bands with significant accuracy.

The spectral angle mapper (SAM) has been widely used in multispectral and hyperspectral image analysis to measure spectral similarity between substance signatures for material identification. It has been shown that the SAM is essentially the Euclidean distance when the spectral angle is small. Most recently, a stochastic measure, called the spectral information divergence (SID), has been suggested to model the spectrum of a hyperspectral image pixel as a probability distribution, so that spectral variations among spectral bands can be captured more effectively in a stochastic manner. This paper develops a new hyperspectral spectral discrimination measure, which combines the SID and the SAM into a mixed measure. More specifically, let r and r denote two hyperspectral image pixel vectors with their corresponding spectra specified by s and s. Then SAM(s,s) measures the spectral angle between s and s. Similarly, SID(s,s) measures the information divergence between the probability distributions generated by s and s. The proposed new measure, referred to as the SID-SAM mixed measure, can be implemented in two versions, given by SID(s,s)×tan(SAM(s,s)) and SID(s,s)×sin(SAM(s,s)), where tan and sin are the usual trigonometric functions. The spectral discriminability of such a mixed measure is greatly enhanced by multiplying the spectral abilities of the two measures. In order to demonstrate its utility, a comparative study is conducted among the SID-SAM mixed measure, the SID, and the SAM. Our experimental results have shown that the discriminatory ability of the (SID,SAM) mixed measure can be a significant improvement over the SID and SAM.

Purpose:Recently, the proportion of patients with high myopia has shown a continuous growing trend, more toward the younger age groups. This study aimed to predict the changes in spherical equivalent refraction (SER) and axial length (AL) in children using machine learning methods.Methods:This study is a retrospective study. The cooperative ophthalmology hospital of this study collected data on 179 sets of childhood myopia examinations. The data collected included AL and SER from grades 1 to 6. This study used the six machine learning models to predict AL and SER based on the data. Six evaluation indicators were used to evaluate the prediction results of the models.Results:For predicting SER in grade 6, grade 5, grade 4, grade 3, and grade 2, the best results were obtained through the multilayer perceptron (MLP) algorithm, MLP algorithm, orthogonal matching pursuit (OMP) algorithm, OMP algorithm, and OMP algorithm, respectively. The R2 of the five models were 0.8997, 0.7839, 0.7177, 0.5118, and 0.1758, respectively. For predicting AL in grade 6, grade 5, grade 4, grade 3, and grade 2, the best results were obtained through the Extra Tree (ET) algorithm, MLP algorithm, kernel ridge (KR) algorithm, KR algorithm, and MLP algorithm, respectively. The R2 of the five models were 0.7546, 0.5456, 0.8755, 0.9072, and 0.8534, respectively.Conclusion:Therefore, in predicting SER, the OMP model performed better than the other models in most experiments. In predicting AL, the KR and MLP models were better than the other models in most experiments.

Examining hand samples can be a necessary step for geologic studies, and effective mapping of such samples can be achieved through the high spectral and spatial resolutions of ground-based hyperspectral imaging (HSI) at the millimeter to centimeter scale. We have developed a simple approach to crude oil identification and characterization — feasible in 16 h — based on hyperspectral data collected under ultraviolet (UV) lighting and normalized with respect to the fluorescence patterns of the Spectralon diffuse reflectance material. The samples under consideration were extracted from a core acquired from an Early Cretaceous bituminous sandstone formation in the Athabasca Basin located near Fort McMurray, Alberta, Canada. This basin contains the largest natural bitumen deposit in the world, where surface mining operations currently are viable only for approximately 20% of the estimated 164 billion barrels of total recoverable oil reserves. This deposit is unique in that its tar sands are water-wet, which facilitates the separation of bitumen from the sandstone via water-based gravity separation. However, large amounts of water are still required for oil recovery; therefore, a fast and reliable way to mark portions of the deposit where ample petroleum has accumulated and assess its extractability based on its physical characteristics prior to mining can be helpful for optimizing resource usage. For this reason, we test and visually develop the ability of three classification methods — the spectral angle mapper, support vector machine, and supervised neural network — to distinguish among bitumen, Spectralon, and a nonfluorescent slate background based on the emission of visible light in response to absorbing UV light of different wavelengths. We also adopt spectral indices useful for indicating concentrated bitumen in tar sands. Errors inherent to the methodology are discussed along with ways to mitigate them. After accounting for these, HSI can be a valuable asset alongside other techniques used for production economics evaluation.

Imaging modalities like hyperspectral imaging create large amounts of data. Time efficient, automated analytic techniques are therefore required to enjoy the power of such methods. In this study it was investigated if hyperspectral imaging followed by automated noise filtering and statistical image analysis is a suitable method for characterization of the macroscopic structure of atherosclerotic lesions. Ten human aorta samples (6×8 cm) were collected during autopsy. Hyperspectral white light and fluorescence images and 5 - 6 biopsies were collected from each sample. The biopsies were stained (HES, Sudan red), and grouped according to histology. All images were noise filtered and normalized. Fluorescence spectra were collected from all biopsied regions, and used to compute average spectra for each histological group. Supervised classification was performed using Spectral angle mapping (SAM) with the average spectra as endmembers. K-means- and ISO-data clustering was used for unsupervised classification. The results show that noise filtering and normalization is essential for reliable classification. Supervised classification was in general found to perform better than unsupervised classification. However, the SAM results strongly depend on the variation in the spectra used to compute the average endmember spectra. The analysis show that fatty deposits, calcifications, connective tissue and hemoglobin can be identified. The lesions were found to have a complex structure where vulnerable regions could be found next to stabile regions within the same lesion. In conclusion hyperspectral imaging, automated filtering and -analysis was found to be a suitable tool to classify advanced atherosclerotic lesions.

Hyperspectral imaging can be a useful remote-sensing technology for classifying tree species. Prior to the image classification stage, effective mapping endeavours must first identify the optimal spectral and spatial resolutions for discriminating the species of interest. Such a procedure may contribute to improving the classification accuracy, as well as the image acquisition planning. In this work, we address the effect of degrading the original bandwidth and pixel size of a hyperspectral and hyperspatial image for the classification of Sclerophyll forest tree species. A HySpex-VNIR 1600 airborne-based hyperspectral image with submetric spatial resolution was acquired in December 2009 for a native forest located in the foothills of the Andes of central Chile. The main tree species of this forest were then sampled in the field between January and February 2010. The original image spectral and spatial resolutions (160 bands with a width of 3.7 nm and pixel sizes of 0.3 m) were systematically degraded by resampling using a Gaussian model and a nearest neighbour method, respectively (until reaching 39 bands with a width of 14.8 nm and pixel sizes of 2.4 m). As a result, 12 images with different spectral and spatial resolution combinations were created. Subsequently, these images were noise-reduced using the minimum noise fraction procedure and 12 additional images were created. Statistical class separabilities from the spectral divergence measure and an assessment of classification accuracy of two supervised hyperspectral classifiers (spectral angle mapper (SAM) and spectral information divergence (SID)) were applied for each of the 24 images. The best overall and per-class classification accuracies (>80%) were observed when the SAM classifier was applied on the noise-reduced reflectance image at its original spectral and spatial resolutions. This result indicates that pixels somewhat smaller than the tree canopy diameters were the most appropriate to represent the spatial variability of the tree species of interest. On the other hand, it suggests that noise-reduced bands derived from the full image spectral resolution rendered the best discrimination of the spectral properties of the tree species of interest. Meanwhile, the better performance of SAM over SID may result from the ability of the former to classify tree species regardless of the illumination differences in the image. This technical approach can be particularly useful in native forest environments, where the irregular surface of the uppermost canopy is subject to a differentiated illumination.