The use of airborne lidar for orchard tree inventory

The tree inventory in orchards is of great interest for orchard management and for government insurance plans. However, the conventional inventory is time‐consuming and expensive. Here a remote sensing method is introduced for orchard inventory. Airborne LIDAR (light detection and ranging) data were employed to obtain tree topography, and multispectral images were used as a reference. LIDAR vector data were converted to raster data for tree crown delineating purpose and in order to be easily superimposed on multispectral data in the same database. A tree crown delineation model was developed using a tree height image derived from the difference between canopy and ground LIDAR altitudes. The number of trees was computed from the delineation model. Spatially separated trees were precisely counted by fine definition of their crowns. For larger trees, although they have irregular crown form, like multi‐tops, holes in the centre or overlapped branches, the model developed in this study provided reliable results for crown delineation.     

A comparison of three methods for automatic tree crown detection and delineation from high spatial resolution imagery

This article compares the performance of three algorithms representative of published methods for tree crown detection and delineation from high spatial resolution imagery, and demonstrates a standardized accuracy assessment framework. The algorithms – watershed segmentation, region growing and valley-following – were tested on softwood and hardwood sites using Emerge natural colour vertical aerial imagery with 60 cm ground sampled distance and QuickBird panchromatic imagery with an 11? look angle. The evaluation considered both plot-level and individual tree crown detection and delineation results. The study shows that while all three methods reasonably delineate crowns in the softwood stand on the Emerge image, region growing provided the highest accuracies, with producer's and user's accuracy for tree detection reaching 70% and root mean square error for crown diameter estimation of 15%. Crown detection accuracies were lower on the QuickBird image. No algorithm proved accurate for the hardwood stand on either image set (both producer's and user's accuracies < 30%).     

Automatic extraction of roads from aerial images based on scale space and snakes

Abstract. We propose a new approach for automatic road extraction from aerial imagery with a model and a strategy mainly based on the multi-scale detection of roads in combination with geometry-constrained edge extraction using snakes. A main advantage of our approach is, that it allows for the first time a bridging of shadows and partially occluded areas using the heavily disturbed evidence in the image. Additionally, it has only few parameters to be adjusted. The road network is constructed after extracting crossings with varying shape and topology. We show the feasibility of the approach not only by presenting reasonable results but also by evaluating them quantitatively based on ground truth. Received: 22 July 1999 / Accepted: 20 March 2000     

Indicators for separating undesirable and well-delineated tree crowns in high spatial resolution images

Much effort has been spent on the automatic detection and delineation of individual trees from high spatial resolution images. However, delineation errors may lead to an inaccurate crown size when compared with ground measurements. Thus, it is problematic to use delineated crowns to derive information on tree variables, e.g. crown diameter, tree height, diameter at breast height (DBH), stand volume, stem volume or stand competition index. In this study, we investigated two indicators – the mean digital number (MDN) within each delineated crown and the difference between MDNs (DMDNs) for 0.6 m buffer zones outside and inside the boundary of each delineated crown – to separate poorly delineated crowns from well-delineated ones. We modelled the relationships between delineated crowns and field-based crown size, between delineated crowns and tree height, and between delineated crowns and DBH observations in a Norway spruce (Picea abies) stand, separately considering models based on all delineated results and crowns identified as being well delineated. Our results showed that the capability of the two indicators in separating poorly and well-delineated crowns varied under different thresholds. The results also indicated that models considering only well-delineated crowns were more robust and effective in estimating and predicting tree crown diameter, DBH and tree height than models that considered all delineated results.     

Automatic extraction and delineation of single trees from remote sensing data

In this paper, we present a novel approach for the automatic extraction of trees and the delineation of the tree crowns from remote sensing data, and report and evaluate the results obtained with different test data sets. The approach is scale-invariant and is based on co-registered colour-infrared aerial imagery and a digital surface model (DSM). Our primary assumption is that the coarse structure of the crown, if represented at the appropriate level in scale-space, can be approximated with the help of an ellipsoid. The fine structure of the crown is suppressed at this scale level and can be ignored. Our approach is based on a tree model with three geometric parameters (size, circularity and convexity of the tree crown) and one radiometric parameter for the tree vitality. The processing strategy comprises three steps. First, we segment a wide range of scale levels of a pre-processed version of the DSM. In the second step, we select the best hypothesis for a crown from the overlapping segments of all levels based on the tree model. The selection is achieved with the help of fuzzy functions for the tree model parameters. Finally, the crown boundary is refined using active contour models (snakes). The approach was tested with four data sets from different sensors and exhibiting different resolutions. The results are very promising and prove the feasibility of the new approach for automatic tree extraction from remote sensing data. The major part of this work was carried out while both authors worked together at the Institute of Photogrammetry and GeoInformation, University of Hannover.     

Comparison of six individual tree crown detection algorithms evaluated under varying forest conditions

In this article, six individual tree crown (ITC) detection/delineation algorithms are evaluated, using an image data set containing six diverse forest types at different geographical locations in three European countries. The algorithms use fundamentally different techniques, including local maxima detection, valley following (VF), region-growing (RG), template matching (TM), scale-space (SS) theory and techniques based on stochastic frameworks. The structurally complexity of the forests in the aerial images used ranges from a homogeneous plantation and an area with isolated tree crowns to an extremely dense deciduous forest type. None of the algorithms alone could successfully analyse all different cases. The study shows that it is important to partition the imagery into homogeneous forest stands prior to the application of individual tree detection algorithms. It furthermore suggests a need for a common, publicly available suite of test images and common test procedures for evaluation of individual tree detection/delineation algorithms. Finally, it highlights that, for complex forest types, monoscopic images are insufficient for consistent tree crown detection, even by human interpreters.     

Quantitative assessment of vegetation structural attributes from aerial photography

Cover of vegetation understorey and overstorey was determined from aerial photography at 1:25 000 and 1:40 000 scales by a grid sampling technique. Models were developed relating values of aerial cover to field cover as determined by intensive field measurement. The influence of photo-scale, photo colour, the angle of the image, shadow, the hiatus between aerial and field sampling, crown width, crown height, proportion of dead trees, drought prior to aerial sampling, land type, previously cleared vegetation and incline on explanatory models was also examined. The only variables that could be clearly interpreted as influencing the models were vegetation height, photo-scale and land type. Only the latter two variables are useful for predictive models. The smaller the scale of photography the greater the exaggeration of the aerial image of tree crowns. This probable result of photo graininess would be most significant when tree crowns are small, an inverse surrogate of tree height. Two-phase models were developed for predicting basal area and biomass from aerial cover. In most instances models were successful for predicting overstorey and understorey cover and for predicting total basal area and biomass. The technique offers a powerful and cost-effective method of assessing vegetation change over long time periods in a way that no other technique can duplicate.     

Classification of Australian forest communities using aerial photography, CASI and HyMap data

Within Australia, the discrimination and mapping of forest communities has traditionally been undertaken at the stand scale using stereo aerial photography. Focusing on mixed species forests in central south-east Queensland, this paper outlines an approach for the generation of tree species maps at the tree crown/cluster level using 1 m spatial resolution Compact Airborne Spectrographic Imager (CASI; 445.8 nm–837.7 nm wavelength) and the use of these to generate stand-level assessments of community composition. Following automated delineation of tree crowns/crown clusters, spectral reflectance from pixels representing maxima or mean-lit averages of channel reflectance or band ratios were extracted for a range of species including Acacia, Angophora, Callitris and Eucalyptus. Based on stepwise discriminant analysis, classification accuracies of dominant species were greatest (87% and 76% for training and testing datasets; n = 398) when the mean-lit spectra associated with a ratio of the reflectance (ρ) at 742 nm (ρ₇₄₂) and 714 nm (ρ₇₁₄) were used. The integration of 2.6 m HyMap (446.1 nm–2477.8 nm) spectra increased the accuracy of classification for some species, largely because of the inclusion of shortwave infrared wavebands. Similar increases in accuracy were achieved when classifications of field spectra resampled to CASI and HyMap wavebands were compared. The discriminant functions were applied subsequently to classify crowns within each image and produce maps of tree species distributions which were equivalent or better than those generated through aerial photograph interpretation. The research provides a new approach to tree species mapping, although some a priori knowledge of the occurrence of broad species groups is required. The tree maps have application to biodiversity assessment in Australian forests.     

Integration of varying spatial,spectral and temporal high-resolution optical images for individual tree crown isolation

Automated methods for capturing geometric and spectral properties of individual tree crowns are becoming increasingly viable options for use in natural resource planning. Crown isolation techniques are needed that are capable of adapting to the changing availability and resolutions of remotely sensed data. Data integration, or the fusion of two distinct data entities, offers a methodological framework that can compensate for the shortcomings of individual datasets while enhancing their desirable features. This study sought to develop a method of data integration for high-resolution optical images of varying spatial and temporal resolutions to improve the automatic detection and delineation of individual tree crowns. A marker-controlled watershed segmentation (MCWS) algorithm was developed for a 30-cm-per-pixel-side airborne colour infrared (CIR) digital image of a leaf-on apple (Malus spp.) orchard. Three methods of obtaining the markers needed for the MCWS algorithm were tested: (1) manual marker selection, (2) template/correlation selection using the 30-cm CIR image, and (3) template/correlation selection using a 15-cm-per-pixel-side true (TRU) colour leaf-off aerial image. The effectiveness of integrating marker data of different temporal and spatial resolutions with the segmentation process of the CIR image scene was tested. A comparison of crown isolation results using markers derived within the segmented 30-cm CIR digital imagery with results from markers derived from the 15-cm TRU image scene indicated greater accuracies to detect and isolate tree crowns with data integration.     

Multipass hierarchical stereo matching for generation of digital terrain models from aerial images

This paper presents a new multi-pass hierarchical stereo-matching approach for generation of digital terrain models (DTMs) from two overlapping aerial images. Our method consists of multiple passes which compute stereo matches with a coarse-to-fine and sparse-to-dense paradigm. An image pyramid is generated and used in the hierarchical stereo matching. Within each pass, the DTM is refined by using the image pyramid from the coarse to the fine level. At the coarsest level of the first pass, a global stereo-matching technique, the intra-/inter-scanline matching method, is used to generate a good initial DTM for the subsequent stereo matching. Thereafter, hierarchical block matching is applied to image locations where features are detected to refine the DTM incrementally. In the first pass, only the feature points near salient edge segments are considered in block matching. In the second pass, all the feature points are considered, and the DTM obtained from the first pass is used as the initial condition for local searching. For the passes after the second pass, 3D interactive manual editing can be incorporated into the automatic DTM refinement process whenever necessary. Experimental results have shown that our method can successfully provide accurate DTM from aerial images. The success of our approach and system has also been demonstrated with a flight simulation software. Received: 4 November 1996 / Accepted: 20 October 1997     

Segmentation of high-resolution InSAR data of a tropical forest using Fourier parameterized deformable models

Currently, tree maps are produced from field measurements that are time consuming and expensive. Application of existing techniques based on aerial photography is often hindered by cloud cover. This has initiated research into the segmentation of high resolution airborne interferometric Synthetic Aperture Radar (SAR) data for deriving tree maps. A robust algorithm is constructed to optimally position closed boundaries. The boundary of a tree crown will be best approximated when at all points on the boundary, the z-coordinate image gradient is maximum, and directed inwards orthogonal to the boundary. This property can be expressed as the result of a line integral along the boundary. Boundaries with a large value for the line integral are likely to be tree crowns. This paper focuses on the search procedure and on illustrating how smoothing can be used to prevent the search from becoming trapped in a local optimum. The final crown detection stage is not described in this paper but could be based on the gradient and implemented using the above described value for the line integral. Results of this paper indicate that a Fourier parametrization with only three harmonics (nine parameters) can describe the shape variation in the 2D crown projection in sufficient detail. Current ground datasets are not suitable for obtaining detection statistics such as the percentage of tree crowns detected and the number of false alarms. Better ground datasets will be needed to evaluate algorithm performance for real tree mapping situations.     

Comparison of three individual tree crown detection methods

Three image processing methods for single tree crown detection in high spatial resolution aerial images are presented and compared using the same image material and reference data. The first method uses templates to find the tree crowns. The other two methods uses region growing. One of them is supported by fuzzy rules while the other uses an image produced by Brownian motion. All three methods detect around 80%, or more, of the visible sunlit trees in two pine Pinus Sylvestris L.) and two spruce stands Picea abies Karst.) in a boreal forest. For all methods, large tree crowns are easier to detect than small ones.     

Estimating stem volume by tree crown area and tree shadow area extracted from pan‐sharpened Quickbird imagery in open Crimean juniper forests

Although open forests represent approximately 30% of the world's forest resources, there is a clear lack of reliable inventory data to allow sustainable management of this valuable resource from semi‐arid areas. This paper demonstrates that the low ground cover of open forest offers a unique opportunity for deriving single tree attributes from high‐resolution satellite imagery, allowing reliable biomass estimation. More particularly, this study investigates the relationship between field‐measured stem volume and tree attributes, including tree crown area and tree shadow area, measured from pan‐sharpened Quickbird imagery with a 0.61 m resolution in a sparse Crimean juniper (Juniperus excelsa M.Bieb.) forest in south‐western Turkey. First tree shadows and crowns were identified and delineated as individual polygons. Both visual delineation and computer‐aided automatic classification methods were tested. After delineation, stem volume as a function of these image‐measured attributes was modelled using linear regression. The statistical analyses indicated that stem volume was correlated with both shadow area and crown area. The best model for stem volume using shadow area resulted in an adjusted R ² = 0.67, with a root mean square error (RMSE) of 12.5%. The model for stem volume using crown area resulted in an adjusted R ² = 0.51, with a RMSE of 15.2%. The results showed that pan‐sharpened Quickbird imagery is suitable for estimating stem volume and may be useful in reducing the time required for obtaining inventory data in open Crimean juniper forests and other similar open forests.     

Global methods for stroke segmentation

Two methods for stroke segmentation from a global point of view are presented and compared. One is based on thinning methods and the other is based on contour curve fitting. For both cases an input image is binarized. For the former, Hilditch's method is used, then crossing points are sought, around which a domain is constructed. Outside the domain, a set of line segments are identified. These lines are connected and approximated by cubic B-spline curves. Smoothly connected lines are selected as segmented curves. This method works well for a limited class of crossing lines, which are shown experimentally. In the latter, a contour line is approximated by cubic B-spline curve, along which curvature is measured. According to the extreme points of the curvature graph, the contour line is segmented, based on which the line segment is obtained. Experimental results are shown for some difficult cases. Received October 31, 1998 / Revised January 12, 1999     

Single tree species classification with a hypothetical multi-spectral satellite

The requirements for high resolution multi-spectral satellite images to be used in single tree species classification for forest inventories are investigated, especially with respect to spatial resolution, sensor noise and geo-registration. In the hypothetical setup, a 3D tree crown map is first obtained from very high resolution panchromatic aerial imagery and subsequently each crown is classified into one of a set of known tree species such that the difference between a model multi-spectral image generated from the 3D crown map and an acquired multi-spectral satellite image of the forested area is minimized. The investigation is conducted partly by generating synthetic data from a 3D crown map from a real mixed forest stand and partly on hypothetical high resolution multi-spectral satellite images obtained from very high resolution colour infrared aerial photographs, allowing different hypothetical spatial resolutions. Conclusions are that until a new generation of even higher resolution satellites becomes available, the most feasible source of remote sensing data for single tree classification will be aerial platforms.     

Single tree detection in heterogeneous boreal forests using airborne laser scanning and area-based stem number estimates

Adaptive single tree detection methods using airborne laser scanning (ALS) data were investigated and validated on 40 large plots sampled from a structurally heterogeneous boreal forest dominated by Norway spruce and Scots pine. Under the working assumption of having uniformly distributed tree locations, area-based stem number estimates were used to guide tree crown delineation from rasterized laser data in two ways: (1) by controlling the amount of smoothing of the canopy height model and (2) by obtaining an appropriate spatial resolution for representing the forest canopy. Single tree crowns were delineated from the canopy height models (CHMs) using a marker-based watershed algorithm, and the delineation results were assessed using a simple tree crown delineation algorithm as a reference method (‘RefMeth’). Using the proposed methods, approximately 46–50% of the total number of trees were detected, while approximately 5–6% false positives were found. The detection rate was, in general, higher for Scots pine than for Norway spruce. The accuracy of individual tree variables (total height and crown width) extracted from the laser data was compared with field-measured data. The individual tree heights were better estimated for deciduous tree species than for the coniferous species Norway spruce and Scots pine. The estimation of crown diameters for Scots pine and deciduous species achieved comparable accuracy, being better than for Norway spruce. The proposed methodology has the potential for easy integration with operational laser scanner-based stand inventories.     

MODEEP: a motion-based object detection and pose estimation method for airborne FLIR sequences

In this paper, we present a method called MODEEP (Motion-based Object DEtection and Estimation of Pose) to detect independently moving objects (IMOs) in forward-looking infrared (FLIR) image sequences taken from an airborne, moving platform. Ego-motion effects are removed through a robust multi-scale affine image registration process. Thereafter, areas with residual motion indicate potential object activity. These areas are detected, refined and selected using a Bayesian classifier. The resulting regions are clustered into pairs such that each pair represents one object's front and rear end. Using motion and scene knowledge, we estimate object pose and establish a region of interest (ROI) for each pair. Edge elements within each ROI are used to segment the convex cover containing the IMO. We show detailed results on real, complex, cluttered and noisy sequences. Moreover, we outline the integration of our fast and robust system into a comprehensive automatic target recognition (ATR) and action classification system.     

A system to navigate a robot into a ship structure

Abstract. A prototype system has been built to navigate a walking robot into a ship structure. The 8-legged robot is equipped with an active stereo head. From the CAD-model of the ship good view points are selected, such that the head can look at locations with sufficient edge features, which are extracted automatically for each view. The pose of the robot is estimated from the features detected by two vision approaches. One approach searches in stereo images for junctions and measures the 3-D position. The other method uses monocular image and tracks 2-D edge features. Robust tracking is achieved with a method of edge projected integration of cues (EPIC). Two inclinometres are used to stabilise the head while the robot moves. The results of the final demonstration to navigate the robot within centimetre accuracy are given.     

Oakwood crown closure estimation by unmixing Landsat TM data

Using an unconstrained least squares solution (LSS) method and an artificial neural network (ANN) algorithm, we estimated oakwood crown closure from a Landsat Thematic Mapper (TM) image of Tulare County, California, USA. Fractions of endmembers (oak crown (f1), grass (f2) and soil (f3)) from mixed pixels were derived from aerial photographs (scale 1?:?40?000) scanned at 1?m ground resolution for training and testing the LSS and ANN algorithms. The aerial photographs were orthorectified using a digital photogrammetric software package with ground control points collected through a differential global positioning system (GPS). The TM image was georeferenced with respect to the corresponding orthorectified aerial photographs. The training and test samples were randomly selected from the TM image and their corresponding fractions of endmembers were derived from the orthophoto. A fourth endmember, shade (f4), was directly extracted from the TM image. Experimental results indicate that the ANN has performed better than the unconstrained LSS. To extract oakwood crown closure in mixed pixels, better results were obtained without using a shade endmember.