José M. Peña

Weed Mapping in Early-Season Maize Fields Using Object-Based Analysis of Unmanned Aerial Vehicle (UAV) Images

The use of remote imagery captured by unmanned aerial vehicles (UAV) has tremendous potential for designing detailed site-specific weed control treatments in early post-emergence, which have not possible previously with conventional airborne or satellite images. A robust and entirely automatic object-based image analysis (OBIA) procedure was developed on a series of UAV images using a six-band multispectral camera (visible and near-infrared range) with the ultimate objective of generating a weed map in an experimental maize field in Spain. The OBIA procedure combines several contextual, hierarchical and object-based features and consists of three consecutive phases: 1) classification of crop rows by application of a dynamic and auto-adaptive classification approach, 2) discrimination of crops and weeds on the basis of their relative positions with reference to the crop rows, and 3) generation of a weed infestation map in a grid structure. The estimation of weed coverage from the image analysis yielded satisfactory results. The relationship of estimated versus observed weed densities had a coefficient of determination of r2=0.89 and a root mean square error of 0.02. A map of three categories of weed coverage was produced with 86% of overall accuracy. In the experimental field, the area free of weeds was 23%, and the area with low weed coverage (<5% weeds) was 47%, which indicated a high potential for reducing herbicide application or other weed operations. The OBIA procedure computes multiple data and statistics derived from the classification outputs, which permits calculation of herbicide requirements and estimation of the overall cost of weed management operations in advance.

An automatic object-based method for optimal thresholding in UAV images

An automatic thresholding algorithm was developed in an OBIA framework.The algorithm was tested in UAV images acquired on different herbaceous row crops.The main objective was to accurately discriminate vegetation vs bare soil.Classification accuracies about 90% were achieved.Two cameras were tested on board the UAV: visible, and visible+infrared. In precision agriculture, detecting the vegetation in herbaceous crops in early season is a first and crucial step prior to addressing further objectives such as counting plants for germination monitoring, or detecting weeds for early season site specific weed management. The ultra-high resolution of UAV images, and the powerful tools provided by the Object Based Image Analysis (OBIA) are the key in achieving this objective. The present research work develops an innovative thresholding OBIA algorithm based on the Otsus method, and studies how the results of this algorithm are affected by the different segmentation parameters (scale, shape and compactness). Along with the general description of the procedure, it was specifically applied for vegetation detection in remotely-sensed images captured with two sensors (a conventional visible camera and a multispectral camera) mounted on an Unmanned Aerial Vehicle (UAV) and acquired over fields of three different herbaceous crops (maize, sunflower and wheat). The tests analyzed the performance of the OBIA algorithm for classifying vegetation coverage as affected by different automatically selected thresholds calculated in the images of two vegetation indices: the Excess Green (ExG) and the Normalized Difference Vegetation Index (NDVI). The segmentation scale parameter affected the vegetation index histograms, which led to changes in the automatic estimation of the optimal threshold value for the vegetation indices. The other parameters involved in the segmentation procedure (i.e., shape and compactness) showed minor influence on the classification accuracy. Increasing the object size, the classification error diminished until an optimum was reached. After this optimal value, increasing object size produced bigger errors.

Object-Based Image Classification of Summer Crops with Machine Learning Methods

The strategic management of agricultural lands involves crop field monitoring each year. Crop discrimination via remote sensing is a complex task, especially if different crops have a similar spectral response and cropping pattern. In such cases, crop identification could be improved by combining object-based image analysis and advanced machine learning methods. In this investigation, we evaluated the C4.5 decision tree, logistic regression (LR), support vector machine (SVM) and multilayer perceptron (MLP) neural network methods, both as single classifiers and combined in a hierarchical classification, for the mapping of nine major summer crops (both woody and herbaceous) from ASTER satellite images captured in two different dates. Each method was built with different combinations of spectral and textural features obtained after the segmentation of the remote images in an object-based framework. As single classifiers, MLP and SVM obtained maximum overall accuracy of 88%, slightly higher than LR (86%) and notably higher than C4.5 (79%). The SVM+SVM classifier (best method) improved these results to 89%. In most cases, the hierarchical classifiers considerably increased the accuracy of the most poorly classified class (minimum sensitivity). The SVM+SVM method offered a significant improvement in classification accuracy for all of the studied crops compared to the conventional decision tree classifier, ranging between 4% for safflower and 29% for corn, which suggests the application of object-based image analysis and advanced machine learning methods in complex crop classification tasks.

High-Throughput 3-D Monitoring of Agricultural-Tree Plantations with Unmanned Aerial Vehicle (UAV) Technology

The geometric features of agricultural trees such as canopy area, tree height and crown volume provide useful information about plantation status and crop production. However, these variables are mostly estimated after a time-consuming and hard field work and applying equations that treat the trees as geometric solids, which produce inconsistent results. As an alternative, this work presents an innovative procedure for computing the 3-dimensional geometric features of individual trees and tree-rows by applying two consecutive phases: 1) generation of Digital Surface Models with Unmanned Aerial Vehicle (UAV) technology and 2) use of object-based image analysis techniques. Our UAV-based procedure produced successful results both in single-tree and in tree-row plantations, reporting up to 97% accuracy on area quantification and minimal deviations compared to in-field estimations of tree heights and crown volumes. The maps generated could be used to understand the linkages between tree grown and field-related factors or to optimize crop management operations in the context of precision agriculture with relevant agro-environmental implications.

Selecting patterns and features for between- and within- crop-row weed mapping using UAV-imagery

The problem of remote weed mapping via machine learning is considered.Unmanned aerial vehicles are used to capture maize and sunflower field images.The proposed method considers pattern and feature selection techniques.The final model requires few user information to generalise to new areas.There are features of great influence for the classification of both crops. This paper approaches the problem of weed mapping for precision agriculture, using imagery provided by Unmanned Aerial Vehicles (UAVs) from sunflower and maize crops. Precision agriculture referred to weed control is mainly based on the design of early post-emergence site-specific control treatments according to weed coverage, where one of the most important challenges is the spectral similarity of crop and weed pixels in early growth stages. Our work tackles this problem in the context of object-based image analysis (OBIA) by means of supervised machine learning methods combined with pattern and feature selection techniques, devising a strategy for alleviating the user intervention in the system while not compromising the accuracy. This work firstly proposes a method for choosing a set of training patterns via clustering techniques so as to consider a representative set of the whole field data spectrum for the classification method. Furthermore, a feature selection method is used to obtain the best discriminating features from a set of several statistics and measures of different nature. Results from this research show that the proposed method for pattern selection is suitable and leads to the construction of robust sets of data. The exploitation of different statistical, spatial and texture metrics represents a new avenue with huge potential for between and within crop-row weed mapping via UAV-imagery and shows good synergy when complemented with OBIA. Finally, there are some measures (specially those linked to vegetation indexes) that are of great influence for weed mapping in both sunflower and maize crops.

Quantifying efficacy and limits of unmanned aerial vehicle (UAV) technology for weed seedling detection as affected by sensor resolution.

In order to optimize the application of herbicides in weed-crop systems, accurate and timely weed maps of the crop-field are required. In this context, this investigation quantified the efficacy and limitations of remote images collected with an unmanned aerial vehicle (UAV) for early detection of weed seedlings. The ability to discriminate weeds was significantly affected by the imagery spectral (type of camera), spatial (flight altitude) and temporal (the date of the study) resolutions. The colour-infrared images captured at 40 m and 50 days after sowing (date 2), when plants had 5–6 true leaves, had the highest weed detection accuracy (up to 91%). At this flight altitude, the images captured before date 2 had slightly better results than the images captured later. However, this trend changed in the visible-light images captured at 60 m and higher, which had notably better results on date 3 (57 days after sowing) because of the larger size of the weed plants. Our results showed the requirements on spectral and spatial resolutions needed to generate a suitable weed map early in the growing season, as well as the best moment for the UAV image acquisition, with the ultimate objective of applying site-specific weed management operations.

An Automatic Random Forest-OBIA Algorithm for Early Weed Mapping between and within Crop Rows Using UAV Imagery

Accurate and timely detection of weeds between and within crop rows in the early growth stage is considered one of the main challenges in site-specific weed management (SSWM). In this context, a robust and innovative automatic object-based image analysis (OBIA) algorithm was developed on Unmanned Aerial Vehicle (UAV) images to design early post-emergence prescription maps. This novel algorithm makes the major contribution. The OBIA algorithm combined Digital Surface Models (DSMs), orthomosaics and machine learning techniques (Random Forest, RF). OBIA-based plant heights were accurately estimated and used as a feature in the automatic sample selection by the RF classifier; this was the second research contribution. RF randomly selected a class balanced training set, obtained the optimum features values and classified the image, requiring no manual training, making this procedure time-efficient and more accurate, since it removes errors due to a subjective manual task. The ability to discriminate weeds was significantly affected by the imagery spatial resolution and weed density, making the use of higher spatial resolution images more suitable. Finally, prescription maps for in-season post-emergence SSWM were created based on the weed maps—the third research contribution—which could help farmers in decision-making to optimize crop management by rationalization of the herbicide application. The short time involved in the process (image capture and analysis) would allow timely weed control during critical periods, crucial for preventing yield loss.

Spatial Quality Evaluation of Resampled Unmanned Aerial Vehicle-Imagery for Weed Mapping

Unmanned aerial vehicles (UAVs) combined with different spectral range sensors are an emerging technology for providing early weed maps for optimizing herbicide applications. Considering that weeds, at very early phenological stages, are similar spectrally and in appearance, three major components are relevant: spatial resolution, type of sensor and classification algorithm. Resampling is a technique to create a new version of an image with a different width and/or height in pixels, and it has been used in satellite imagery with different spatial and temporal resolutions. In this paper, the efficiency of resampled-images (RS-images) created from real UAV-images (UAV-images; the UAVs were equipped with two types of sensors, i.e., visible and visible plus near-infrared spectra) captured at different altitudes is examined to test the quality of the RS-image output. The performance of the object-based-image-analysis (OBIA) implemented for the early weed mapping using different weed thresholds was also evaluated. Our results showed that resampling accurately extracted the spectral values from high spatial resolution UAV-images at an altitude of 30 m and the RS-image data at altitudes of 60 and 100 m, was able to provide accurate weed cover and herbicide application maps compared with UAV-images from real flights.

Machine learning paradigms for weed mapping via unmanned aerial vehicles

This paper presents a novel strategy for weed monitoring, using images taken with unmanned aerial vehicles (UAVs) and concepts of image analysis and machine learning. Weed control in precision agriculture designs site-specific treatments based on the coverage of weeds, where the key is to provide precise weed maps timely. Most traditional remote platforms, e.g. piloted planes or satellites, are, however, not suitable for early weed monitoring, given their low temporal and spatial resolutions, as opposed to he ultra-high spatial resolution of UAVs. The system here proposed makes use of UAV-imagery and is based on: 1) Divide the image, 2) compute and binarise the vegetation indexes, 3) detect crop rows, 4) optimise the parameters and 4) learn a classification model. Since crops are usually organised in rows, the use of a crop row detection algorithm helps to separate properly weed and crop pixels, which is a common handicap given the spectral similitude of both. Several artificial intelligence paradigms are compared in this paper to identify the most suitable strategy for this topic (i.e. unsupervised, supervised and semi-supervised approaches). Our experiments also study the effect of different parameteres: the flight altitude, the sensor and the use of previously trained models at a different height. Our results show that 1) very promising performance can be obtained, even when using very few labelled data and 2) the classification model can be learnt in a subplot of the experimental field at low altitude and then applied to the whole field at a higher height, which simplifies the whole process. These results motivate the use of this strategy to design weed monitoring strategies for early post-emergence weed control.

An Experimental Comparison for the Identification of Weeds in Sunflower Crops via Unmanned Aerial Vehicles and Object-Based Analysis

Weed control in precision agriculture refers to the design of site-specific control treatments according to weed coverage and it is very useful to minimise costs and environmental risks. The crucial component is to provide precise and timely weed maps via weed monitoring. This paper compares different approaches for weed mapping using imagery from Unmanned Aerial Vehicles in sunflower crops. We explore different alternatives, such as object-based analysis, which is a strategy that is spreading rapidly in the field of remote sensing. The usefulness of these approaches is tested by considering support vector machines, one of the most popular machine learning classifiers. The results show that the object-based analysis is more promising than the pixel-based one and demonstrate that both the features related to vegetation indexes and those related to the shape of the objects are meaningful for the problem.