Multispectral Algorithm Research Articles

Multispectral Texture Benchmark

Dozens of textural features have been published, but their realistic validation for efficient recognition applications still needs to be discovered. Textural features are derived using various approaches. We present a benchmark that can be used to evaluate these features and categorize them based on their information efficiency. We propose how the features can be benchmarked and explain different ways of measuring their properties and performance. Most textural feature-extracting algorithms are only based on information extraction from monospectral images (gray-level). Apart from native multispectral algorithms, we generalize some of these originally monospectral features for hyperspectral textures in our illustrating examples.

Dark Channel and Cross Channel Based Multi-Prior Combined Multi-Spectral Super-Resolution Algorithm

Dark Channel and Cross Channel Based Multi-Prior Combined Multi-Spectral Super-Resolution Algorithm

Advancing cyanobacteria biomass estimation from hyperspectral observations: Demonstrations with HICO and PRISMA imagery

Retrieval of the phycocyanin concentration (PC), a characteristic pigment of, and proxy for, cyanobacteria biomass, from hyperspectral satellite remote sensing measurements is challenging due to uncertainties in the remote sensing reflectance (∆Rrs) resulting from atmospheric correction and instrument radiometric noise. Although several individual algorithms have been proven to capture local variations in cyanobacteria biomass in specific regions, their performance has not been assessed on hyperspectral images from satellite sensors. Our work leverages a machine-learning model, Mixture Density Networks (MDNs), trained on a large (N = 939) dataset of collocated in situ chlorophyll-a concentrations (Chla), PCs, and remote sensing reflectance (Rrs) measurements to estimate PC from all relevant spectral bands. The performance of the developed model is demonstrated via PC maps produced from select images of the Hyperspectral Imager for the Coastal Ocean (HICO) and Italian Space Agency's PRecursore IperSpettrale della Missione Applicativa (PRISMA) using a matchup dataset. As input to the MDN, we incorporate a combination of widely used band ratios (BRs) and line heights (LHs) taken from existing multispectral algorithms, that have been proven for both Chla and PC estimation, as well as novel BRs and LHs to increase the overall cyanobacteria biomass estimation accuracy and reduce the sensitivity to ∆Rrs. When trained on a random half of the dataset, the MDN achieves uncertainties of 44.3%, which is less than half of the uncertainties of all viable optimized multispectral PC algorithms. The MDN is notably better than multispectral algorithms at preventing overestimation on low (<10 mg m−3) PC. Visibly, HICO and PRISMA PC maps show the wider dynamic range that can be represented by the MDN. The available in situ and satellite-derived Rrs matchups and measured in situ PC demonstrate the robustness of the MDN for estimating low (<10 mg m−3) PC and the reduced impact of ∆Rrs on medium-to-high in situ PC (>10 mg m−3). According to our extensive assessments, the developed model is anticipated to enable practical PC products from PRISMA and HICO, therefore the model is promising for planned hyperspectral missions, such as the Plankton Aerosol and Cloud Ecosystem (PACE). This advancement will enhance the complementary roles of hyperspectral radiometry from satellite and low-altitude platforms for quantifying and monitoring cyanobacteria harmful algal blooms at both large and local spatial scales.

Una propuesta cronológica para la Cueva de El Castillo (Puente Viesgo, Cantabria), basada en su estratigrafía iconográfica

Some years ago, we began a review of the work done by E. Ripoll in 1953. During this time, we have examined not only each and every one of the figures discovered by him, but have also expanded the list, reaching nearly all the 450 new figures in the entire cave. This significant increase is thanks to the use of new technologies, such as a 3D scanner, digital filters, multispectral algorithms, giga images, or the newest innovation, the use, for the first time in our studies on cave art, of hyperspectral methodology. The latter gives us access to a much wider light spectrum than the one created by multispectral analysis.

Edge-guided multispectral image fusion algorithm

Most existing multispectral fusion algorithms often suffer from spectral or spatial information distortion. Driven by this motivation, we propose an edge-guided multispectral (MS) image fusion algorithm. In particular, it combines the advantages of generative adversarial networks and improved fusion frameworks, so the merged image can better preserve the spectral information of the original multispectral image while injecting spatial detail information. Specifically, first, an MS image with more image detail is generated using the generated confrontation network for preliminary reconstruction. The panchromatic image edge information and the antagonistic learning strategy are introduced for the robust multispectral image reconstruction. Then, using the reconstructed MS image and the general component substitution image fusion framework, the whole fusion system of this paper is constructed. An enhancement operator is introduced to inject spatial details. Our extensive dataset evaluations show that our approach performs better in terms of high objective quality and human visual perception than several of the most advanced fusion methods.

基于偏最小二乘法的多光谱降维算法

基于偏最小二乘法的多光谱降维算法

Automatic optic disc detection in colour fundus images by means of multispectral analysis and information content.

The optic disc (OD) in retinal fundus images is widely used as a reference in computer-based systems for the measurement of the severity of retinal disease. A number of algorithms have been published in the past 5 years to locate and measure the OD in digital fundus images. Our proposed algorithm, automatically: (i) uses the three channels (RGB) of the digital colour image to locate the region of interest (ROI) where the OD lies, (ii) measures the Shannon information content per channel in the ROI, to decide which channel is most appropriate for searching for the OD centre using the circular Hough transform. A series of evaluations were undertaken to test our hypothesis that using the three channels gives a better performance than a single channel. Three different databases were used for evaluation purposes with a total of 2,371 colour images giving a misdetection error of 3% in the localisation of the centre of the OD. We find that the area determined by our algorithm which assumes that the OD is circular, is similar to that found by other algorithms that detected the shape of the OD. Five metrics were measured for comparison with other recent studies. Combining the two databases where expert delineation of the OD is available (1,240 images), the average results for our multispectral algorithm are: TPR = 0.879, FPR = 0.003, Accuracy = 0.994, Overlap = 80.6% and Dice index = 0.878.

Rapid Detection and Visualization of Mechanical Bruises on “Nanfeng” Mandarin Using the Hyperspectral Imaging Combined with ICA_LSQ Method

The hyperspectral imaging (HSI) is used in this work to detect slight mechanical bruises (indentation) on “Nanfeng” mandarin (NM) surface. The independent component analysis (ICA) and the least square method (LSM) are used for data dimension reduction and determination of the coefficients corresponding to the images at characteristic wavelengths, respectively. The Lambertian model and the mean normalization method are used for spectral data preprocessing before applying the ICA, and then, six images centered at the wavelengths of 556 nm, 578 nm, 593 nm, 604 nm, 616 nm, and 683 nm are selected by analyzing the separating matrix of the ICA4. Subsequently, the LSM is used to calculate the coefficients of the ICA4 to enhance the visual contrast of a combination image between the indentation and sound region on the fruit surface. During detecting the indentation on a fruit, the stem-end and calyx of the fruit can be misclassified as indentation, so another six images centered at wavelengths of 556 nm, 604 nm, 647 nm, 670 nm, 716 nm, and 819 nm are selected based on the ICA2 to segment the stem-end and calyx before detecting an indentation region. The experimental results show that performance of the image segmentation algorithm is acceptable, the total success rate for indentation and sound fruits is 93.98%, and the total success rate for stem-end and calyx is 100%, indicating that the proposed multispectral algorithm is capable of detecting the indentation on fruit. The images at the characteristic wavelengths can also be used to establish an online multispectral imaging system for detecting the indentation regions on fruits.

A rational, multispectral mapping algorithm for primary motor cortex: A primary step before cortical stimulation.

For future artificial intelligence-based brain mapping, development of a rational and safe scoring system for a brain motor mapping algorithm using electrocorticography (ECoG score), which contains various spectral, purely intrinsic brain activities, is necessary for either before or in the absence of electrical cortical stimulation (ECS). We evaluated 1114 electrodes of 10 consecutive focal epilepsy patients who underwent subdural electrode implantation before epilepsy surgery at Kyoto University Hospital during 2011-2017. Data from ECoG-based mapping (bandpass filter of 0.016-300/600Hz) to define the primary motor area (M1) localization were used to create an ECoG score (range = 0-4) by assigning 1 point each for the occurrence of ECoG components: very slow movement-related cortical potentials (<0.5-1.0Hz), event-related synchronization (76-100Hz or 100-200Hz), and event-related desynchronization (8-12Hz or 12-24Hz). The ECoG score was assessed by calculating the sensitivity, specificity, and cutoff values of the score for localization concordance with M1 defined using only ECS as a reference. With an area under the receiver operating characteristic curve (AUC) of 0.76, cutoffs of scores of 4 and 1 showed high specificity (94%) and sensitivity (98%) in concordance with ECS-based mapping, respectively. The ECoG score for mapping M1 of the upper limb achieved greater accuracy (AUC=0.85) compared to that of the face (AUC=0.64). The ECoG score proposed in the present study is rational, simple, and useful to define M1, and it is spatially concordant with ECS. Although ECS is still widely employed for presurgical examination, our proposed application of the ECoG score may be suitable for future brain M1 mapping, and possibly beyond M1 mapping, independently of ECS.

Evaluation and assimilation of various satellite-derived rainfall products over India

ABSTRACT Accurate prediction of rainfall from the numerical weather prediction model is one of the major objectives over tropical regions. In this study, four different satellite-derived rainfall products (viz. merged-rainfall product from TRMM (Tropical Rainfall Measuring Mission) 3B42 and IMERG (Integrated Multi-satellitE Retrievals for GPM (Global Precipitation Measurement)), and Indian meteorological satellite INSAT-3D retrieved HEM (Hydro-Estimator Method) and IMSRA (INSAT Multi-Spectral Rainfall Algorithm) rainfall) are assimilated in the Weather Research and Forecasting (WRF) model using variational method. Before assimilation of satellite retrieved rainfall product in the WRF model, selected rainfall products are compared with ground rainfall from India Meteorological Department during Indian summer monsoon (June–September) 2015. Preliminary validation results show root-mean-square-difference (mean difference) of 18.1 (2.1), 21.3 (2.1), 15.4 (−0.72), and 14.4 (0.5) mm day−1 in IMSRA, HEM, IMERG, and TRMM 3B42 rainfall, respectively. Further, the four-dimensional variational data assimilation method is used daily to assimilate selected rainfall products in the WRF model during the entire month of August 2015. Results suggest that assimilation of satellite rainfall improved the WRF model analyses and subsequent temperature and moisture forecasts. Moreover, rainfall prediction is also improved with the maximum positive impact from TRMM rainfall assimilation followed by IMERG rainfall assimilation. Similar nature of improvements is also seen in rainfall prediction when INSAT-3D retrieved rainfall products (HEM and IMSRA) are used for assimilation.

Integrated research of a multi-wavelength method in anisotropic scattering flame on soot temperature and radiative coefficient reconstruction.

At present, the studies of scattering flame problems combined with multi-wavelength reconstruction technology are rather limited. In this paper, a multi-wavelength method combined with a collocation spectral method (CSM)-discrete ordinates method (DOM) forward method is developed for the simultaneous reconstruction of temperature and inhomogeneous radiative coefficient of an axisymmetric laminar ethylene diffusion flame. The scattering source term in the radiative transfer equation is solved by the forward solution method based on DOM, which couples a spatial scheme of CSM and the spherical rings arithmetic progression quadrature. The Tikhonov regularized method is employed to overcome the ill-posed matrix to obtain the global radiative source term. The boundary radiation intensity is input into the algorithm as a known value for inversion of temperature, absorption coefficient, and scattering coefficient. The retrieval results demonstrate that the temperature and the radiation parameters of the scattering flame can be well reconstructed for the flame with axisymmetric distribution, even for the flame with noisy data. In addition, the scattering coefficient is more difficult to rebuild than temperature. The inversion accuracy does not deteriorate with a slight increase in noise. Also, the addition of scattering terms to the proposed algorithm can improve the accuracy in reconstructing the radiation parameters. The developed method in this paper would be useful for multispectral algorithms, which could be used in later experiments based on hyperspectral techniques.

Aerosol Optical Thickness (AOT) Assessment Using GIS &amp; Remote Sensing

Atmospheric aerosol particles are one of the significant agents of air quality degradation. MODIS, GIS and Remote Sensing techniques have made the way of AOT assessment easiest over historic manual systems. This paper concerns itself with the AOT assessment using GIS and Remote Sensing over Dhaka, Bangladesh in 2017. Required observations for AOT assessment are taken considering seasonal variability. Considered three seasons for this research are winter (January, February and December), Pre-monsoon (March and April) and Post-monsoon (October and November). Monsoon variations are not considered to avoid excessive cloud correction. The multispectral algorithm model is used to detect AOT considering the surface condition homogenous. Basically, Landsat 8+ OLI images are used for AOT assessment and NASA Earth Observation (NEO) data are used for data validation. It is found that the winter season has the highest concentration of AOT compared to pre and post monsoon. This is due to the meteorological factors like cloud, rainfall, humidity, wind pressure and speed, temperature etc.

Shortwave IR Adaption of the Mid-Infrared Radiance Method of Fire Radiative Power (FRP) Retrieval for Assessing Industrial Gas Flaring Output

The radiative power (MW) output of a gas flare is a useful metric from which the rate of methane combustion and carbon dioxide emission can be inferred for inventorying purposes and regular global surveys based on such assessments are now being used to keep track of global gas flare reduction efforts. Several multispectral remote sensing techniques to estimate gas flare radiative power output have been developed for use in such surveys and single band approaches similar to those long used for the estimation of landscape fire radiative power output (FRP) can also be applied. The MIR-Radiance method, now used for FRP retrieval within the MODIS active fire products, is one such single band approach—but its applicability to gas flare targets (which are significantly hotter than vegetation fires) has not yet been assessed. Here we show that the MIR-Radiance approach is in fact not immediately suitable for retrieval of gas flare FRP due to their higher combustion temperatures but that switching to use data from a SWIR (rather than MWIR) spectral channel once again enables the method to deliver unbiased FRP retrievals. Over an assumed flaring temperature range of 1600–2200 K we find a maximum FRP error of ±13.6% when using SWIR observations at 1.6 µm and ±6.3% when using observations made at 2.2 µm. Comparing these retrievals to those made by the multispectral VIIRS ‘NightFire’ algorithm (based on Planck Function fits to the multispectral signals) we find excellent agreement (bias = 0.5 MW, scatter = 1.6 MW). An important implication of the availability of this new SWIR radiance method for gas flare analysis is the potential to apply it to long time-series from older and/or more spectrally limited instruments, unsuited to the use of multispectral algorithms. This includes the ATSR series of sensors operating between 1991–2012 on the ERS-1, ERS-2 and ENVISAT satellites and such long-term data can be used with the SWIR-Radiance method to identify key trends in global gas flaring that have occurred over the last few decades.

Aerosol Optical Thickness (AOT) Assessment Using GIS &amp;amp; Remote Sensing

Atmospheric aerosol particles are one of the significant agents of air quality degradation. MODIS, GIS and Remote Sensing techniques have made the way of AOT assessment easiest over historic manual systems. This paper concerns itself with the AOT assessment using GIS and Remote Sensing over Dhaka, Bangladesh in 2017. Required observations for AOT assessment are taken considering seasonal variability. Considered three seasons for this research are winter (January, February and December), Pre-monsoon (March and April) and Post-monsoon (October and November). Monsoon variations are not considered to avoid excessive cloud correction. The multispectral algorithm model is used to detect AOT considering the surface condition homogenous. Basically, Landsat 8+ OLI images are used for AOT assessment and NASA Earth Observation (NEO) data are used for data validation. It is found that the winter season has the highest concentration of AOT compared to pre and post monsoon. This is due to the meteorological factors like cloud, rainfall, humidity, wind pressure and speed, temperature etc.

On the use of temporal vegetation indices in support of eligibility controls for EU aids in agriculture

ABSTRACTThe use of remote sensing in the context of the Common Agricultural Policy (CAP) has progressively become an official method to support European (EU) Member States in carrying out controls about declarations of farmers requiring EU subsidies in agriculture.Reliable automatic or semi-automatic methodologies aiming at crop identification are still being developed and the only technique, which is officially accepted in the CAP context, remains photo interpretation of high/very high (satellite or aerial) orthoimages. To verify past situations, only orthophotos can be used but, unfortunately, they are not always available. In these cases, the use of satellite sensors with adequate spatial, spectral, and temporal resolutions, together with a reliable data analysis technique, could support or even substitute orthophoto interpretation.In this study, we propose a multi-temporal, multispectral algorithm exploiting the Thematic Mapper/Enhanced Thematic Mapper Plus data on Landsat platforms to identify different land covers in the context of CAP. Here it is presented to discriminate arable from non-arable lands. Assessment of the methodology was carried out using Corine 2012 and more than 1500 validation points over Basilicata region (Southern Italy). A general good agreement was found (74%), which increases to 82% in the specific case of arable land identification.

Detection of Asian Dust Storm Using MODIS Measurements

Every year, a large number of aerosols are released from dust storms into the atmosphere, which may have potential impacts on the climate, environment, and air quality. Detecting dust aerosols and monitoring their movements and evolutions in a timely manner is a very significant task. Satellite remote sensing has been demonstrated as an effective means for observing dust aerosols. In this paper, an algorithm based on the multi-spectral technique for detecting dust aerosols was developed by combining measurements of moderate resolution imaging spectroradiometer (MODIS) reflective solar bands and thermal emissive bands. Data from dust events that occurred during the past several years were collected as training data for spectral and statistical analyses. According to the spectral curves of various scene types, a series of spectral bands was selected individually or jointly, and corresponding thresholds were defined for step-by-step scene classification. The multi-spectral algorithm was applied mainly to detect dust storms in Asia. The detection results were validated not only visually with MODIS true color images, but also quantitatively with products of Ozone Monitoring Instrument (OMI) and Cloud Aerosol Lidar with Orthogonal Polarization (CALIOP). The validations showed that this multi-spectral detection algorithm was suitable to monitor dust aerosols in the selected study areas.

Estimating Air Particulate Matter 10 Using Landsat Multi-Temporal Data and Analyzing its Annual Temporal Pattern over Gaza Strip, Palestine

Palestine faces economic and political issues such as conflict, siege and urbanization; all these have led to an increase in the air pollution over Gaza Strip. In this study, Particulate matter 10 (PM10) concentrations over Gaza Strip has been estimated by Landsat Thematic Mapper (TM) and Landsat Enhanced Thematic Mapper Plus (ETM+) data, based on a multispectral algorithm. Simultaneously, in-situ measurements for the corresponding particulate are acquired for selected time period. Landsat and ground data for eleven years are used to develop the algorithm while four years data (2002, 2006, 2010 and 2014) have been used to validate the results of algorithm. The developed algorithm gives highest regression, R coefficient value i.e. 0.86; RMSE value as 9.71 µg/m³; P values as 0. Average validation of algorithm show that calculated PM10 strongly correlates with measured PM10, indicating high efficiency of algorithm for the mapping of PM10 concentration during the years 2000 to 2014. Overall results show increase in minimum, maximum and average yearly PM10 concentrations, also presents similar trend over urban area. The rate of urbanization has been evaluated by supervised classification of the Landsat image. Urban sprawl from year 2000 to 2014 results in a high concentration of PM10 in the study area.

Fast detection and visualization of early decay in citrus using Vis-NIR hyperspectral imaging

Early detection of fungal infection in citrus fruit is one of the major problems in the postharvest phase. The automation of this task is still a challenge for the citrus industry. In this study, the potential application of hyperspectral imaging, which combines conventional imaging and spectroscopy to simultaneously acquire both spatial and spectral information from an object, was evaluated for automatic detection of the early symptoms of decay caused by Penicillium digitatum fungus in citrus fruit. Hyperspectral images of sound and decayed navel oranges were acquired in the wavelength range of 325–1100nm. Principal component analysis (PCA) was applied to a dataset comparing of average spectra from decayed and sound tissue to reduce the dimensionality of data and to observe the ability of visible-near infrared (Vis-NIR) hyper-spectra to discriminate data from two classes. And, a mean normalization step is applied prior to PCA to reduce the effect of sample curvature on spectral profiles. In this case it was observed that sound and decayed spectra were separable along the resultant first principal component (PC1) axis, then, four wavelength images centered at 575, 698, 810 and 969nm were selected as the characteristic wavelength images by analyzing the weight coefficients of PC1 in order to develop a fast classification method for establishing an on-line multispectral imaging system. Subsequently, a combination image, which obtained by multiplying the characteristic weight coefficients by corresponding to mean-normalized characteristic wavelength images of each orange sample, was calculated for determination of decayed fruits. Based on the obtained multispectral combination image, the technique of intensity slicing as one of the pseudo-color image processing methods is used to transform the combination image into a 2-D visual classification image that would enhance the contrast between sound and decayed classes. Finally, an image segmentation algorithm for detection of decayed fruit was developed based on the pseudo-color image coupled with a simple thresholding method. For the investigated 210 naval orange samples including 80 sound fruits and 130 infected fruits, the total success rate is 100% for training set and 98.6% for test set with no false negatives, respectively, indicating that the proposed multispectral algorithm here is capable of detecting decay caused by penicillium digitatum in naval orange fruit using only four key wavelength images. The results from this study could be used for development of a non-destructive monitoring system for rapid detection of decayed citrus on the processing line. The idea behind the proposed algorithm can be extended to detect the non-visible damages of other fruit, such as slight bruise and chilling injury in apples.

Detection of Lettuce Discoloration Using Hyperspectral Reflectance Imaging.

Rapid visible/near-infrared (VNIR) hyperspectral imaging methods, employing both a single waveband algorithm and multi-spectral algorithms, were developed in order to discrimination between sound and discolored lettuce. Reflectance spectra for sound and discolored lettuce surfaces were extracted from hyperspectral reflectance images obtained in the 400–1000 nm wavelength range. The optimal wavebands for discriminating between discolored and sound lettuce surfaces were determined using one-way analysis of variance. Multi-spectral imaging algorithms developed using ratio and subtraction functions resulted in enhanced classification accuracy of above 99.9% for discolored and sound areas on both adaxial and abaxial lettuce surfaces. Ratio imaging (RI) and subtraction imaging (SI) algorithms at wavelengths of 552/701 nm and 557–701 nm, respectively, exhibited better classification performances compared to results obtained for all possible two-waveband combinations. These results suggest that hyperspectral reflectance imaging techniques can potentially be used to discriminate between discolored and sound fresh-cut lettuce.

Multispectral detection of skin defects of bi-colored peaches based on vis–NIR hyperspectral imaging

Fruit skin defects may cause fruit spoilage, reduce commodity economic value, and give rise to food quality and safety concerns. Therefore, one of the main tasks of post-harvest processing of fruit is the detection of skin defects by machine vision technology. However, inspection of skin defects on bi-colored fruit varieties by image processing is more difficult because of the high variability of the skin color. This article presents a multispectral detection method for skin defects of bi-colored ‘Pinggu’ peaches based on visible-near infrared (vis–NIR) hyperspectral imaging. Peaches with nine types of skin condition including skin injury, scarring, insect damage, puncture injury, decay, disease spots, dehiscent scarring and anthracnose and normal surface were studied. Principal component analysis (PCA) was used to reduce hyperspectral data dimensionality to select several wavelengths that could potentially be used in an in-line multispectral imaging system. Different defect types produced an obvious feature only in some specific PC images depending on whether the visible light spectrum (425–780nm), the near infrared spectrum (781–1000nm), the full-spectrum (400–1000nm) or only characteristic wavelengths (463, 555, 687, 712, 813, 970nm or 781, 815, 848nm) were used. A two-band ratio image (Q781/848) was successfully used to differentiate defects from a normal surface. Finally, a detection algorithm for skin defects was developed based on a band ratio (Q781/848) coupled with a simple thresholding method. For the investigated 145 independent test samples with nine skin conditions, an accuracy of 96.6% was obtained, indicating that the proposed multispectral algorithm was effective in differentiating normal and defective bi-colored peaches. The proposed algorithm can be extended to other fruit.