Tree species classification with hyperspectral imaging and lidar

Rudjord, Oystein; Trier, Øivind Due

doi:10.1109/whispers.2016.8071665

Cited by 5 publications

(3 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Rudjord and Trier (2016) proposed to compare average spectral profiles of spruce, pine and birch to locate a few wavelengths that may be used in ratios to separate the tree species. As noted earlier, the scaled radiance spectra ( Figure 5) are better suited than the original spectra (Figure 4) for this analysis.…”

Section: Classification Methodsmentioning

confidence: 99%

Tree species classification in Norway from airborne hyperspectral and airborne laser scanning data

Trier

Salberg

Kermit

et al. 2018

European Journal of Remote Sensing

View full text Add to dashboard Cite

This article compares four new automatic methods to discriminate between spruce, pine and birch, which are the dominating tree species in Norwegian forests. Airborne laser scanning and hyperspectral data were used. The laser scanning data was used to mask pixels with low or no vegetation in the hyperspectral data. A green-blue ratio was used to remove shadow areas from tree canopies, and the normalized difference vegetation index to remove dead vegetation and non-vegetation. The best method was hyperspectral pixel classification with 160 spectral channels in the visible and near-infrared spectrum, using a deep neural network. This method achieved 87% correct classification rate. Partial least squares regression for hyperspectral pixel classification achieved 78%. Deep neural network image classification using canopy height blended with three hyperspectral channels achieved 74%. A simple pixel classification method based on two spectral indices resulted in 67% correct classification. A possible future improvement is to find a better way to combine hyperspectral data with canopy height data in a deep neural network. ARTICLE HISTORY

show abstract

Section: Classification Methodsmentioning

confidence: 99%

Tree species classification in Norway from airborne hyperspectral and airborne laser scanning data

Trier

Salberg

Kermit

et al. 2018

European Journal of Remote Sensing

View full text Add to dashboard Cite

show abstract

“…In this multi-level morphological active contour algorithm, LiDAR pictures were suggested as the input for the description and delineation of the tree. Oystein Rudjord and Oivind Due Trier et.al [8] This article introduces a fresh technique for distinguishing spruce, pine, and birch, the dominant tree species in Norwegian forests. To this end, ALS and hyperspectral information are used concurrently.…”

Section: Introductionmentioning

confidence: 99%

Clustering of Trees from Panchromatic Images

Thotakura,

Koneru,

Kollu

2019

IJEAT

View full text Add to dashboard Cite

Tree Clustering from satellite images assists in ecological environmental protection. It also helps in managing green resources to provide sustainable development guidance. The automatic clustering of trees is a challenging task. Many models tend to give poor results when there is noise in the image. The aim is to propose a model for clustering of tree crown from panchromatic satellite image using image processing algorithms. In the proposed model we use Cartosat-2 satellite data and the image data is pre-processed to enhance the resulted image analyzed using segmentation models. The resulted image is trained using the clustering model which classifies the tree crowns from the panchromatic images. The proposed model can be able to classify tree crowns effectively from satellite imagery. The proposed model also calculates the tree crown height and width from satellite imagery.

show abstract

“…En la actualidad, las cámaras espectrales han demostrado ser una herramienta eficaz en la industria alimentaria para el control de adulteraciones [78,79] y para la evaluación de la calidad y seguridad de productos alimentarios [80,81], también en el ámbito sanitario [82,83] y en el estudio de la Tierra y medioambiente [84,85]. Otro sector en el que las cámaras espectrales destacan especialmente es el agrícola, donde se emplean, por ejemplo, en la agricultura de precisión [86,87], control de fitopatologías [88,89] y discriminación de especies o variedades [90,91],…”

unclassified

Desarrollo de modelos para la identificación de variedades y fitopatologías mediante técnicas fotónicas

Borraz-Martínez¹

View full text Add to dashboard Cite

In the cultivation of fruit trees, the most common is to work with cuttings. In this way, the grafted fruit tree variety offers specific fruit qualities and the rootstock offers certain agronomic and adaptation characteristics. The grafts that come from varieties of the same species usually present a great genetic proximity, which makes them very difficult to identify by means of agro-morphological characters. For this reason, it is necessary to implement more complex methods to be able to distinguish them. Nowadays, the most widely used methods for varietal identification are based on DNA analysis, which allow obtaining molecular profiles. However, these biomolecular techniques require sophisticated instruments and facilities, time consuming, and are expensive. These drawbacks make the implementation of molecular techniques unfeasible as methods for the routine analysis of large volumes of samples. In this context, the nursery plant sector demands a rapid, no-destructive, accurate, and economic methods. The main objective of this thesis is to develop a method that allows identifying and differentiating varieties at the nursery plant level, together with the ability to assess their physiological status. The thesis is articulated in 4 articles published in indexed journals together with essays presented in the present work. The main results are to a) Protocols have been developed for the use of near infrared spectroscopy techniques, hyperspectral camera and artificial vision coupled to deep learning in fruit cuttings and cherry rootstocks; b) Near-infrared spectroscopy proved to be a technique capable of differentiating, with a very high degree of specificity, six varieties of almond (Prunus dulcis), but it presented limitations for its implementation in a production line; c) The use of the hyperspectral camera has proven adequate for the differentiation of three rootstock varieties and the analysis of the physiological state, including the premature detection of Pseudomonas syringae; d) The use of artificial vision coupled with deep learning has proven to be effective in the differentiation of almond varieties; e) The joint analysis of the results indicates that the best solution to be implemented in a production line would go through the implementation of the spectral camera for the control of the physiological state of the plants and the artificial vision coupled to deep learning for the varietal identification. En el cultivo de árboles frutales lo más habitual es trabajar con plantas injertadas. De esta forma, la variedad de árbol frutal injertada ofrece unas cualidades de fruto específicas y el portainjerto ofrece unas características agronómicas y de adaptación determinadas. Los injertos que provienen de variedades de una misma especie suelen presentar una gran proximidad genética, lo que genera que sean muy difíciles de identificar mediante caracteres agromorfológicos. Por esta razón, es necesario la implementación de métodos más complejos para poder distinguirlos. En la actualidad, los métodos más utilizados para la identificación varietal se basan en el análisis del ADN, los cuales permiten obtener perfiles moleculares. Sin embargo, estas técnicas biomoleculares requieren de instrumentos e instalaciones sofisticadas, no son rápidas de ejecutar y presentan un coste elevado. Estos inconvenientes hacen inviable la implementación de las técnicas moleculares como métodos de análisis rutinario de grandes volúmenes de muestras. En este contexto, el sector viverístico demanda nuevos métodos que seas rápidos, no destructivos, eficaces y con un coste por análisis bajo. El objetivo principal de esta tesis es desarrollar un método que permita identificar y diferenciar variedades a nivel de planta de vivero junto con la capacidad de evaluar el estado fisiológico de las mismas. La tesis se articula en 4 artículos publicados en revistas indexadas junto con otros ensayos expuestos en el presente trabajo. Los principales resultados son: a) Se han desarrollado protocolos para el uso de técnicas de espectroscopia de infrarrojo cercano, cámara hiperespectral y visión artificial acoplado a deep learning en esquejes de frutales y portainjertos de cerezo; b) La espectroscopia de infrarrojo cercano demostró ser una técnica capaz de diferenciar, con un grado muy elevado de especificidad, seis variedades de almendro (Prunus dulcis), pero presentó limitaciones para su implementación en una línea de producción; c) El uso de la cámara hiperespectrales ha demostrado adecuado para la diferenciación de tres variedades de portainjerto y el análisis del estado fisiológico, incluyendo la detección prematura de Pseudomonas syringae; d) El uso de la visión artificial acoplada a deep learning ha demostrado ser eficaz en la diferenciación de variedades de almendro; e) El análisis conjunto de los resultados indican que la mejor solución para ser implementada en una línea de producción pasaría por la implementación de la cámara espectral para el control del estado fisiológico de las plantas y la visión artificial acoplada a deep learning para la identificación varietal.

show abstract

Tree species classification with hyperspectral imaging and lidar

Cited by 5 publications

References 3 publications

Tree species classification in Norway from airborne hyperspectral and airborne laser scanning data

Tree species classification in Norway from airborne hyperspectral and airborne laser scanning data

Clustering of Trees from Panchromatic Images

Desarrollo de modelos para la identificación de variedades y fitopatologías mediante técnicas fotónicas

Contact Info

Product

Resources

About