Assessment of residual slash coverage using UAVs and implications for aspen regeneration

2020 ◽  
Vol 8 (1) ◽  
pp. 19-29
Author(s):  
Landon L. Sealey ◽  
Ken C.J. Van Rees
Keyword(s):  
Maximum Likelihood ◽  
High Resolution ◽  
Image Classification ◽  
Populus Tremuloides ◽  
Classification Accuracy ◽  
Forest Floor ◽  
Aerial Imagery ◽  
Detailed Method ◽  
Slash Distribution ◽  
Supervised Image Classification

Proper redistribution of residual slash following harvesting is crucial for ensuring successful regeneration and continued health in trembling aspen (Populus tremuloides) forests. As traditional methods of measuring residual slash are a strenuous and tedious process, the objective of this study was to develop a new, faster, and more detailed method to assess residual slash distribution for entire harvested blocks. This study also aimed to assess the influence residual slash coverage had on the success of aspen regeneration 1 year after winter harvesting. Using high-resolution UAV imagery and maximum likelihood supervised image classification, residual slash was differentiated from the underlying forest floor. Overall, classification accuracy ranged between 85% and 96% with the highest accuracy occurring when aerial imagery was collected at the beginning of the second spring following winter harvesting. Slash distribution was quite consistent across harvested blocks, with 92% of harvested blocks experiencing <33% coverage. There was no relationship between the level of aspen regeneration following 1 year of growth and percentage slash coverage up to 60%. No vegetation plots occurred in areas with >60% slash coverage; therefore, it is unknown whether aspen regeneration will be affected in areas with higher slash coverage.

scholarly journals The Influence of Region of Interest Heterogeneity on Classification Accuracy in Wetland Systems

2019 ◽  
Vol 11 (5) ◽  
pp. 551 ◽  
Author(s):  
Tedros M. Berhane ◽  
Hugo Costa ◽  
Charles R. Lane ◽  
Oleg A. Anenkhonov ◽  
Victor V. Chepinoga ◽  
...  
Keyword(s):  
Image Classification ◽  
Classification Accuracy ◽  
Region Of Interest ◽  
Complex Environments ◽  
Sampling Frame ◽  
Single Class ◽  
Natural Systems ◽  
Data Points ◽  
Large Heterogeneity ◽  
Supervised Image Classification

Classifying and mapping natural systems such as wetlands using remote sensing frequently relies on data derived from regions of interest (ROIs), often acquired during field campaigns. ROIs tend to be heterogeneous in complex systems with a variety of land cover classes. However, traditional supervised image classification is predicated on pure single-class observations to train a classifier. This ultimately encourages end-users to create single-class ROIs, nudging ROIs away from field-based points or gerrymandering the ROI, which may produce ROIs unrepresentative of the landscape and potentially insert error into the classification. In this study, we explored WorldView-2 images and 228 field-based data points to define ROIs of varying heterogeneity levels in terms of class membership to classify and map 22 discrete classes in a large and complex wetland system. The goal was to include rather than avoid ROI heterogeneity and assess its impact on classification accuracy. Parametric and nonparametric classifiers were tested with ROI heterogeneity that varied from 7% to 100%. Heterogeneity was governed by ROI area, which we increased from the field-sampling frame of ~100 m2 nearly 19-fold to ~2124 m2. In general, overall accuracy (OA) tended downwards with increasing heterogeneity but stayed relatively high until extreme heterogeneity levels were reached. Moreover, the differences in OA were not statistically significant across several small-to-large heterogeneity levels. Per-class user’s and producer’s accuracies behaved similarly. Our findings suggest that ROI heterogeneity did not harm classification accuracy unless heterogeneity became extreme, and thus there are substantial practical advantages to accommodating heterogeneous ROIs in image classification. Rather than attempting to avoid ROI heterogeneity by gerrymandering, classification in wetland environments, as well as analyses of other complex environments, should embrace ROI heterogeneity.

scholarly journals CLASS-BASED AFFINITY PROPAGATION FOR HYPERSPECTRAL IMAGE DIMENSIONALITY REDUCTION AND IMPROVEMENT OF MAXIMUM LIKELIHOOD CLASSIFICATION ACCURACY

2019 ◽  
Vol 25 (1) ◽  
Author(s):  
André Moiane ◽  
Alvaro Muriel Lima Machado
Keyword(s):  
Feature Selection ◽  
Maximum Likelihood ◽  
Dimensionality Reduction ◽  
Image Classification ◽  
Classification Accuracy ◽  
Clustering Algorithm ◽  
Infrared Imaging ◽  
Hyperspectral Image ◽  
Spectral Feature ◽  
Affinity Propagation

Abstract This paper investigates an alternative classification method that integrates class-based affinity propagation (CAP) clustering algorithm and maximum likelihood classifier (MLC) with the purpose of overcome the MLC limitations in the classification of high dimensionality data, and thus improve its accuracy. The new classifier was named CAP-MLC, and comprises two approaches, spectral feature selection and image classification. CAP clustering algorithm was used to perform the image dimensionality reduction and feature selection while the MLC was employed for image classification. The performance of MLC in terms of classification accuracy and processing time is determined as a function of the selection rate achieved in the CAP clustering stage. The performance of CAP-MLC has been evaluated and validated using two hyperspectral scenes from the Airborne Visible Infrared Imaging Spectrometer (AVIRIS) and the Hyperspectral Digital Imagery Collection Experiment (HYDICE). Classification results show that CAP-MLC observed an enormous improvement in accuracy, reaching 94.15% and 96.47% respectively for AVIRIS and HYDICE if compared with MLC, which had 85.42% and 81.50%. These values obtained by CAP-MLC improved the MLC classification accuracy in 8.73% and 14.97% for these images. The results also show that CAP-MLC performed well, even for classes with limited training samples, surpassing the limitations of MLC.

scholarly journals Deep Semi-Supervised Image Classification Algorithms: a Survey

2021 ◽  
Vol 27 (12) ◽  
pp. 1390-1407
Author(s):  
Ani Vanyan ◽  
Hrant Khachatrian
Keyword(s):  
Machine Learning ◽  
Image Classification ◽  
Supervised Learning ◽  
Classification Accuracy ◽  
Classification Algorithms ◽  
The Past ◽  
Supervised Learning Algorithms ◽  
Supervised Image Classification ◽  
Learning Focused ◽  
Remarkable Progress

Semi-supervised learning is a branch of machine learning focused on improving the performance of models when the labeled data is scarce, but there is access to large number of unlabeled examples. Over the past five years there has been a remarkable progress in designing algorithms which are able to get reasonable image classification accuracy having access to the labels for only 0.1% of the samples. In this survey, we describe most of the recently proposed deep semi-supervised learning algorithms for image classification and identify the main trends of research in the field. Next, we compare several components of the algorithms, discuss the challenges of reproducing the results in this area, and highlight recently proposed applications of the methods originally developed for semi-supervised learning.

scholarly journals Assessing Riyadh’s Urban Change Utilizing High-Resolution Imagery

Land ◽  
2019 ◽  
Vol 8 (12) ◽  
pp. 193
Author(s):  
Ali Alghamdi ◽  
Anthony R. Cummings
Keyword(s):  
Maximum Likelihood ◽  
High Resolution ◽  
Accuracy Assessment ◽  
Middle Eastern ◽  
Object Oriented ◽  
Training Data ◽  
High Resolution Imagery ◽  
Significant Research ◽  
Desert Areas ◽  
Spot 5

The implications of change on local processes have attracted significant research interest in recent times. In urban settings, green spaces and forests have attracted much attention. Here, we present an assessment of change within the predominantly desert Middle Eastern city of Riyadh, an understudied setting. We utilized high-resolution SPOT 5 data and two classification techniques—maximum likelihood classification and object-oriented classification—to study the changes in Riyadh between 2004 and 2014. Imagery classification was completed with training data obtained from the SPOT 5 dataset, and an accuracy assessment was completed through a combination of field surveys and an application developed in ESRI Survey 123 tool. The Survey 123 tool allowed residents of Riyadh to present their views on land cover for the 2004 and 2014 imagery. Our analysis showed that soil or ‘desert’ areas were converted to roads and buildings to accommodate for Riyadh’s rapidly growing population. The object-oriented classifier provided higher overall accuracy than the maximum likelihood classifier (74.71% and 73.79% vs. 92.36% and 90.77% for 2004 and 2014). Our work provides insights into the changes within a desert environment and establishes a foundation for understanding change in this understudied setting.

Sign in / Sign up

Export Citation Format

Share Document