Automated delineation of individual tree crowns in high spatial resolution aerial images by multiple-scale analysis

1998 ◽  
Vol 11 (2) ◽  
pp. 64-73 ◽  
Author(s):  
Tomas Brandtberg ◽  
Fredrik Walter
Keyword(s):  
Spatial Resolution ◽  
High Spatial Resolution ◽  
Multiple Scale ◽  
Aerial Images ◽  
Scale Analysis ◽  
Individual Tree ◽  
Tree Crowns ◽  
Multiple Scale Analysis

Automatic individual tree based analysis of high spatial resolution aerial images on naturally regenerated boreal forests

1999 ◽  
Vol 29 (10) ◽  
pp. 1464-1478 ◽  
Author(s):  
Tomas Brandtberg
Keyword(s):  
Spatial Resolution ◽  
Boreal Forests ◽  
High Spatial Resolution ◽  
Ground Truth ◽  
Aerial Images ◽  
Individual Tree ◽  
Ground Truth Data ◽  
European Aspen ◽  
Rms Error ◽  
Complementary Image

Individual tree based forest surveys are feasible using modern computer technology. The presented approach for analysing high spatial resolution (pixel size 10 cm) aerial images of naturally regenerated boreal forests is based on visible significant trees. Sunlight patches on the ground are suppressed, followed by optimal image smoothing. The problem with inclined illumination is handled by adapted thresholding. Each connected threshold segment (a collection of one or more trees) is further smoothed. A selection of the resulting convex edge segments is used for identifying significant tree crown circles. Six complementary image variables are estimated and used for regression analysis. An evaluation of the ground-truth data in central Sweden gives good results on the stem position estimate (a root mean square (RMS) error of 108 cm) and the stem number estimate (a relative RMS error of 11%). The complementary variables contribute significantly to the stem diameter prediction, resulting in the following experimental values: Scots pine (Pinus sylvestris L.) (R2 = 59.5%, s = 4.9 cm, N = 157), Norway spruce (Picea abies (L.) Karst.) (R2 = 21.9%, s = 6.4 cm, N = 398), birch (Betula pubescens Ehrh.) (R2 = 35.4%, s = 5.3 cm, N = 133), and European aspen (Populus tremula L.) (R2 = 61.4%, s = 4.6 cm, N = 13). The results indicate strong species dependence.

scholarly journals Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information

2021 ◽  
Vol 13 (3) ◽  
pp. 364
Author(s):  
Han Gao ◽  
Jinhui Guo ◽  
Peng Guo ◽  
Xiuwan Chen
Keyword(s):  
Deep Learning ◽  
Land Cover ◽  
Spatial Resolution ◽  
Large Scale ◽  
High Spatial Resolution ◽  
Training Data ◽  
Aerial Images ◽  
Rural Landscapes ◽  
Feature Representations ◽  
Object Based

Recently, deep learning has become the most innovative trend for a variety of high-spatial-resolution remote sensing imaging applications. However, large-scale land cover classification via traditional convolutional neural networks (CNNs) with sliding windows is computationally expensive and produces coarse results. Additionally, although such supervised learning approaches have performed well, collecting and annotating datasets for every task are extremely laborious, especially for those fully supervised cases where the pixel-level ground-truth labels are dense. In this work, we propose a new object-oriented deep learning framework that leverages residual networks with different depths to learn adjacent feature representations by embedding a multibranch architecture in the deep learning pipeline. The idea is to exploit limited training data at different neighboring scales to make a tradeoff between weak semantics and strong feature representations for operational land cover mapping tasks. We draw from established geographic object-based image analysis (GEOBIA) as an auxiliary module to reduce the computational burden of spatial reasoning and optimize the classification boundaries. We evaluated the proposed approach on two subdecimeter-resolution datasets involving both urban and rural landscapes. It presented better classification accuracy (88.9%) compared to traditional object-based deep learning methods and achieves an excellent inference time (11.3 s/ha).

scholarly journals A Comparison of Methods for Determining Forest Composition from High-Spatial-Resolution Remotely Sensed Imagery

Forests ◽  
2021 ◽  
Vol 12 (9) ◽  
pp. 1290
Author(s):  
Benjamin T. Fraser ◽  
Russell G. Congalton
Keyword(s):  
Spatial Resolution ◽  
High Spatial Resolution ◽  
Remotely Sensed ◽  
Local Scale ◽  
Google Earth ◽  
Forest Composition ◽  
Visual Interpretation ◽  
Individual Tree ◽  
Remotely Sensed Imagery ◽  
Composition Classes

Remotely sensed imagery has been used to support forest ecology and management for decades. In modern times, the propagation of high-spatial-resolution image analysis techniques and automated workflows have further strengthened this synergy, leading to the inquiry into more complex, local-scale, ecosystem characteristics. To appropriately inform decisions in forestry ecology and management, the most reliable and efficient methods should be adopted. For this reason, our research compares visual interpretation to digital (automated) processing for forest plot composition and individual tree identification. During this investigation, we qualitatively and quantitatively evaluated the process of classifying species groups within complex, mixed-species forests in New England. This analysis included a comparison of three high-resolution remotely sensed imagery sources: Google Earth, National Agriculture Imagery Program (NAIP) imagery, and unmanned aerial system (UAS) imagery. We discovered that, although the level of detail afforded by the UAS imagery spatial resolution (3.02 cm average pixel size) improved the visual interpretation results (7.87–9.59%), the highest thematic accuracy was still only 54.44% for the generalized composition groups. Our qualitative analysis of the uncertainty for visually interpreting different composition classes revealed the persistence of mislabeled hardwood compositions (including an early successional class) and an inability to consistently differentiate between ‘pure’ and ‘mixed’ stands. The results of digitally classifying the same forest compositions produced a higher level of accuracy for both detecting individual trees (93.9%) and labeling them (59.62–70.48%) using machine learning algorithms including classification and regression trees, random forest, and support vector machines. These results indicate that digital, automated, classification produced an increase in overall accuracy of 16.04% over visual interpretation for generalized forest composition classes. Other studies, which incorporate multitemporal, multispectral, or data fusion approaches provide evidence for further widening this gap. Further refinement of the methods for individual tree detection, delineation, and classification should be developed for structurally and compositionally complex forests to supplement the critical deficiency in local-scale forest information around the world.

Multiple scale analysis of factors influencing the distribution of an invasive aquatic grass

2008 ◽  
Vol 11 (8) ◽  
pp. 1903-1912 ◽  
Author(s):  
Sarina E. Loo ◽  
Ralph Mac Nally ◽  
Dennis J. O’Dowd ◽  
James R. Thomson ◽  
P. S. Lake
Keyword(s):  
Multiple Scale ◽  
Scale Analysis ◽  
Factors Influencing ◽  
Multiple Scale Analysis

scholarly journals Experiment and Theory on the Nonlinear Vibration of a Shallow Arch Under Harmonic Excitation at the End

2005 ◽  
Vol 74 (6) ◽  
pp. 1061-1070 ◽  
Author(s):  
Jen-San Chen ◽  
Cheng-Han Yang
Keyword(s):  
Steady State ◽  
Excitation Frequency ◽  
Multiple Scale ◽  
Steady State Response ◽  
Scale Analysis ◽  
Jump Phenomenon ◽  
Shallow Arch ◽  
State Response ◽  
Geometrical Imperfection ◽  
Multiple Scale Analysis

In this paper we study, both theoretically and experimentally, the nonlinear vibration of a shallow arch with one end attached to an electro-mechanical shaker. In the experiment we generate harmonic magnetic force on the central core of the shaker by controlling the electric current flowing into the shaker. The end motion of the arch is in general not harmonic, especially when the amplitude of lateral vibration is large. In the case when the excitation frequency is close to the nth natural frequency of the arch, we found that geometrical imperfection is the key for the nth mode to be excited. Analytical formula relating the amplitude of the steady state response and the geometrical imperfection can be derived via a multiple scale analysis. In the case when the excitation frequency is close to two times of the nth natural frequency two stable steady state responses can exist simultaneously. As a consequence jump phenomenon is observed when the excitation frequency sweeps upward. The effect of geometrical imperfection on the steady state response is minimal in this case. The multiple scale analysis not only predicts the amplitudes and phases of both the stable and unstable solutions, but also predicts analytically the frequency at which jump phenomenon occurs.

Sign in / Sign up

Export Citation Format

Share Document