Michele Dalponte
University of Trento
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Michele Dalponte.
IEEE Transactions on Geoscience and Remote Sensing | 2008
Michele Dalponte; Lorenzo Bruzzone; Damiano Gianelle
In this paper, we propose an analysis on the joint effect of hyperspectral and light detection and ranging (LIDAR) data for the classification of complex forest areas. In greater detail, we present: 1) an advanced system for the joint use of hyperspectral and LIDAR data in complex classification problems; 2) an investigation on the effectiveness of the very promising support vector machines (SVMs) and Gaussian maximum likelihood with leave-one-out-covariance algorithm classifiers for the analysis of complex forest scenarios characterized from a high number of species in a multisource framework; and 3) an analysis on the effectiveness of different LIDAR returns and channels (elevation and intensity) for increasing the classification accuracy obtained with hyperspectral images, particularly in relation to the discrimination of very similar classes. Several experiments carried out on a complex forest area in Italy provide interesting conclusions on the effectiveness and potentialities of the joint use of hyperspectral and LIDAR data and on the accuracy of the different classification techniques analyzed in the proposed system. In particular, the elevation channel of the first LIDAR return was very effective for the separation of species with similar spectral signatures but different mean heights, and the SVM classifier proved to be very robust and accurate in the exploitation of the considered multisource data.
IEEE Transactions on Geoscience and Remote Sensing | 2013
Michele Dalponte; Hans Ole Ørka; Terje Gobakken; Damiano Gianelle; Erik Næsset
Tree species mapping in forest areas is an important topic in forest inventory. In recent years, several studies have been carried out using different types of hyperspectral sensors under various forest conditions. The aim of this work was to evaluate the potential of two high spectral and spatial resolution hyperspectral sensors (HySpex-VNIR 1600 and HySpex-SWIR 320i), operating at different wavelengths, for tree species classification of boreal forests. To address this objective, many experiments were carried out, taking into consideration: 1) three classifiers (support vector machines (SVM), random forest (RF), and Gaussian maximum likelihood); 2) two spatial resolutions (1.5 m and 0.4 m pixel sizes); 3) two subsets of spectral bands (all and a selection); and 4) two spatial levels (pixel and tree levels). The study area is characterized by the presence of four classes 1) Norway spruce, 2) Scots pine, together with 3) scattered Birch and 4) other broadleaves. Our results showed that: 1) the HySpex VNIR 1600 sensor is effective in boreal tree species classification with kappa accuracies over 0.8 (with Pine and Spruce reaching producers accuracies higher than 95%); 2) the role of the HySpex-SWIR 320i is limited, and its bands alone are able to properly separate only Pine and Spruce species; 3) the spatial resolution has a strong effect on the classification accuracy (an overall decrease of more than 20% between 0.4 m and 1.5 m spatial resolution); and 4) there is no significant difference between SVM or RF classifiers.
IEEE Transactions on Geoscience and Remote Sensing | 2011
Michele Dalponte; Lorenzo Bruzzone; Damiano Gianelle
Forest inventories are important tools for the management of forests. In this context, the estimation of the tree stem volume is a key issue. In this paper, we present a system for the estimation of forest stem diameter and volume at individual tree level from multireturn light detection and ranging (LIDAR) data. The proposed system is made up of a preprocessing module, a LIDAR segmentation algorithm (aimed at retrieving tree crowns), a variable extraction and selection procedure, and an estimation module based on support vector regression (SVR) (which is compared with a multiple linear regression technique). The variables derived from LIDAR data are computed from both the intensity and elevation channels of all available returns. Three different methods of variable selection are analyzed, and the sets of variables selected are used in the estimation phase. The stem volume is estimated with two methods: 1) direct estimation from the LIDAR variables and 2) combination of diameters and heights estimated from LIDAR variables with the species information derived from a classification map according to standard height/diameter relationships. Experimental results show that the system proposed is effective and provides high accuracies in both the stem volume and diameter estimations. Moreover, this paper provides useful indications on the effectiveness of SVR with LIDAR in forestry problems.
International Journal of Applied Earth Observation and Geoinformation | 2013
Alessandro Montaghi; Piermaria Corona; Michele Dalponte; Damiano Gianelle; Gherardo Chirici; Håkan Olsson
Abstract This article reviews the recent literature concerning airborne laser scanning for forestry purposes in Italy, and presents the current methodologies used to extract forest characteristics from discrete return ALS (Airborne Laser Scanning) data. Increasing interest in ALS data is currently being shown, especially for remote sensing-based forest inventories in Italy; the driving force for this interest is the possibility of reducing costs and providing more accurate and efficient estimation of forest characteristics. This review covers a period of approximately ten years, from the first application of laser scanning for forestry purposes in 2003 to the present day, and shows that there are numerous ongoing research activities which use these technologies for the assessment of forest attributes (e.g., number of trees, mean tree height, stem volume) and ecological issues (e.g., gap identification, fuel model detection). The basic approaches – such as single tree detection and area-based modeling – have been widely examined and commented in order to explore the trend of methods in these technologies, including their applicability and performance. Finally this paper outlines and comments some of the common problems encountered in operational use of laser scanning in Italy, offering potentially useful guidelines and solutions for other countries with similar conditions, under a rather variable environmental framework comprising Alpine, temperate and Mediterranean forest ecosystems.
Scandinavian Journal of Forest Research | 2013
Hans Ole Ørka; Michele Dalponte; Terje Gobakken; Erik Næsset; Liviu Theodor Ene
Abstract The purpose of the study was to evaluate tree species composition estimated using combinations of different remotely sensed data with different inventory approaches for a forested area in Norway. Basal area species composition was estimated as both species proportions and main species by using data from airborne laser scanning (ALS) and airborne (multispectral and hyperspectral) imagery as auxiliary information in combination with three different inventory approaches: individual tree crown (ITC) approach; semi-individual tree crown (SITC) approach; and area-based approach (ABA). The main tree species classification obtained an overall accuracy higher than 86% for all ABA alternatives and for the two other inventory approaches (ITC and SITC) when combining ALS and hyperspectral imagery. The correlation between estimated species proportions and species proportions measured in the field was higher for coniferous species than for deciduous species and increased with the spectral resolution used. Especially, the ITC approach provided more accurate information regarding the proportion of deciduous species that occurred only in small proportions in the study area. Furthermore, the species proportion estimates of 83% of the plots deviated from field measured species proportions by two-tenths or less. Thus, species composition could be accurately estimated using the different approaches and the highest levels of accuracy were attained when ALS was used in combination with hyperspectral imagery. The accuracies obtained using the ABA in combination with only ALS data were encouraging for implementation in operational forest inventories.
Brain and Language | 2011
Brian Murphy; Massimo Poesio; Francesca Bovolo; Lorenzo Bruzzone; Michele Dalponte; Heba Lakany
Achieving a clearer picture of categorial distinctions in the brain is essential for our understanding of the conceptual lexicon, but much more fine-grained investigations are required in order for this evidence to contribute to lexical research. Here we present a collection of advanced data-mining techniques that allows the category of individual concepts to be decoded from single trials of EEG data. Neural activity was recorded while participants silently named images of mammals and tools, and category could be detected in single trials with an accuracy well above chance, both when considering data from single participants, and when group-training across participants. By aggregating across all trials, single concepts could be correctly assigned to their category with an accuracy of 98%. The pattern of classifications made by the algorithm confirmed that the neural patterns identified are due to conceptual category, and not any of a series of processing-related confounds. The time intervals, frequency bands and scalp locations that proved most informative for prediction permit physiological interpretation: the widespread activation shortly after appearance of the stimulus (from 100 ms) is consistent both with accounts of multi-pass processing, and distributed representations of categories. These methods provide an alternative to fMRI for fine-grained, large-scale investigations of the conceptual lexicon.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing | 2009
Michele Dalponte; Lorenzo Bruzzone; Damiano Gianelle
Small footprint Light Detection and Ranging (LiDAR) data have been shown to be a very accurate technology to predict stem volume. In particular, most recent sensors are able to acquire multiple return (more than 2) data at very high hit density, allowing one to have detailed characterization of the canopy. In this paper, we utilize very high density ( >8 hits per m2) LiDAR data acquired over a forest stand in Italy. Our approach was as follows: Individual trees were first extracted from the LiDAR data and a series of attributes from both the first, and non-first (multiple), hits associated with each crown were then extracted. These variables were then correlated with ground truth individual estimates of stem volume. Our results indicate that: (i) non-first returns are informative for the estimation of stem volume (in particular the second return); (ii) some attributes (e.g., maximum at the power of n) better emphasize the information content of returns different from the first respect to other metrics (e.g., minimum, mean); and (iii) the combined use of variables belonging to different returns slightly increases the overall model accuracy. Moreover, we found that the best model for stem volume estimation (adj - R2 = 0.77, P < 0.0001, SE = 0.06) comprised four variables belonging to three returns (first, second, and third). The results of this analysis are important as they underline the effectiveness of the use of multiple return LiDAR data, underling the connection between LiDAR hits different from the first and tree structure and characteristics.
Archive | 2014
Jari Vauhkonen; Hans Ole Ørka; Johan Holmgren; Michele Dalponte; Johannes Heinzel; Barbara Koch
Species-specific information is important for many tasks related to forest management. We review the use of airborne laser scanning (ALS) and complementary data for providing this information. The main ALS-based information is related to structural features, intensity of the echoes, and waveform parameters, whereas spectral information may be provided by fusing data from different sensors. Various types of classifiers are applied, the current emphasis being in non-linear or otherwise complex techniques. The results are successful with respect to the main species, whereas the overall accuracy depends on the desired level of detail in the classification. We expect fusion approaches combining ALS and especially hyperspectral data to become more common and further improvements by the development of advanced sensor technology.
European Journal of Remote Sensing | 2015
Michele Dalponte; Francesco Reyes; Kaja Kandare; Damiano Gianelle
Abstract In this paper four different delineation methods based on airborne laser scanning (ALS) and hyperspectral data are compared over a forest area in the Italian Alps. The comparison was carried out in terms of detected trees, while the ALS based methods are compared also in terms of attributes estimated (e.g. height). From the experimental results emerged that ALS methods outperformed hyperspectral one in terms of tree detection rate in two of three cases. The best results were achieved with a method based on region growing on an ALS image, and by one based on clustering of raw ALS point cloud. Regarding the estimates of the tree attributes all the ALS methods provided good results with very high accuracies when considering only big trees.
IEEE Transactions on Geoscience and Remote Sensing | 2014
Claudio Persello; Abdeslam Boularias; Michele Dalponte; Terje Gobakken; Erik Næsset; Bernhard Schölkopf
Active learning typically aims at minimizing the number of labeled samples to be included in the training set to reach a certain level of classification accuracy. Standard methods do not usually take into account the real annotation procedures and implicitly assume that all samples require the same effort to be labeled. Here, we consider the case where the cost associated with the annotation of a given sample depends on the previously labeled samples. In general, this is the case when annotating a queried sample is an action that changes the state of a dynamic system, and the cost is a function of the state of the system. In order to minimize the total annotation cost, the active sample selection problem is addressed in the framework of a Markov decision process, which allows one to plan the next labeling action on the basis of an expected long-term cumulative reward. This framework allows us to address the problem of optimizing the collection of labeled samples by field surveys for the classification of remote sensing data. The proposed method is applied to the ground sample collection for tree species classification using airborne hyperspectral images. Experiments carried out in the context of a real case study on forest inventory show the effectiveness of the proposed method.