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

Significance Statement

Efficient and timely post-emergence weed control based on herbicide use is a crucial task in crop production because a wrong early weed management not only reduces the crop yield (due to weed competition) but also increases the negative impacts of herbicides on the environment.

Inappropriate weed management is often related with two main problems. The first is applying herbicides when weeds are not in the suitable phenological stage (generally the correct growth stage is when weeds have 4-6 true leaves, although this slightly depends on specific weed species or group of species), the second is broadcasting herbicides over the whole field, even when weed-free areas are present due to the widely demonstrated weed patchy distribution. The first problem is usually addressed using the expert knowledge of farmers. The other problem can be overcome by developing site-specific weed management (SSWM) strategies according to weed emergence. These strategies may consist of both a single herbicide treatment to weed patches where a unique group of weeds is present (for example either grass or broadleaved weeds), or use of several herbicides according to the presence of different weed species or group composition, such as grass, broadleaved weeds or a specific problematic weed (i.e., a herbicide resistant weed).

To determine the weed emergence, it is necessary to generate the weed cover maps, which allow the translation of the spatial distribution of the weed infestation into site-specific herbicide treatment maps. Remote sensing using imagery from an unmanned aerial vehicle (UAV) is now the principal source to monitor weeds at very early phenological stage in a cost effective way due they can carry (even simultaneously) different sensors to register information at diverse spectral ranges, fly at different altitudes to adjust the desired high spatial resolution, and be programmed on demand at key stages of the crops. This is critical when detecting weeds in crops for early post-emergence SSWM when crops and weeds are at the same early phenological stage and they show spectral and visual similarities.

This article is a part of a wide research included in an overall program to investigate the prospects and limitations of UAV imagery in the accurate mapping of a set of problematic and perennial weeds for sustainable and precise use of herbicides by covering a range of technological and methodological key components. Summarizing the following aspects were developed and evaluated:

i) the versatility of the configuration of onboard sensors,

ii) flight altitude,

iii) flight planning,

iv) the percentage of forward (lateral) and side (longitudinal) overlapped imagery and the corresponding ortho-rectification to create an accurately geo-referenced orthomosaicked image of the entire plot for further classification,

v) the development of object-based-image-analysis for an early weed detection emerged between the crop-rows, and finally

vi) the selection of patterns and features not only for a between but also for a within crop-row early weed detection and mapping.

The results clearly demonstrate the feasibility of the UAV-orthomosaicked imagery for both the early detection and mapping of weeds in crops and the saving of herbicides (up to 70% depending on the weed infestation). These findings thus are very useful for designing a field program for the site-specific herbicide applications in the entire field at very early post-emergence. This also provides a base for a future study focused on a number of specific and frequent field conditions, such as the early detection of weeds in curved crop rows.

This global work was supported by the projects AGL2014-52465-C4-4-R (Spanish MINECO, EU-FEDER funds) and RECUPERA-2020 (an agreement between CSIC
and Spanish MINECO, EU-FEDER funds).  

 Selecting patterns and features for between- and within- crop-row weed mapping using UAV-imagery (Advances in Engineering)

Figure credit: Dr. Francisca López-Granados, CSIC, Spanish National Research Council

About the author

María Pérez-Ortiz was born in Cordoba (Spain). She received the B.S. degree in Computer Science (2011) and the M.Sc. degree in Intelligent Systems (2012) from the University of Córdoba, Spain. In 2015, she obtained the Ph.D. degree in Computer Science and Artificial Intelligence in the University of Córdoba, for which she has received 4 awards.

She is now working as a lecturer with the Department of Quantitative Methods at the Universidad Loyola Andalucía (Spain). Her current interests include a wide range of topics concerning machine learning and pattern recognition. 

About the author

José M. Peña was born in Jaén (Spain). He received the B.S. degree in Agricultural Engineering (2000) and the Ph.D. degree in Agricultural Sciences (2006), both from the University of Córdoba (Spain). He developed a postdoctoral stay at the Universidad de California-Davis (USA, 2008-2010) working mainly in the applications of remote sensing for crop and environmental modeling. He is currently a “Ramón y Cajal” Assistant Researcher at the Spanish National Research Council (CSIC).

His current research interests include remote sensing and geospatial technologies, crop mapping, precision agriculture and weed science. 

About the author

Pedro-Antonio Gutiérrez was born in Córdoba (Spain). He received the B.S. degree in Computer Science from the University of Seville (Spain, 2006), and the Ph.D. degree in Computer Science and Artificial Intelligence from the University of Granada (Spain, 2009). He developed two post-doctoral stays at University of Birmingham (United Kingdom, 2009; 2011). He is currently an Assistant Professor with the Department of Computer Science and Numerical Analysis, University of Córdoba (Spain).

His current research interests include pattern recognition, evolutionary computation, and their applications in Precision Agriculture, Human Health and Economy  (AYRNA Group: http://www.uco.es/grupos/ayrna/index.php). 

About the author

Jorge Torres-Sánchez was born in Córdoba (Spain). He received the B.S. degree in Forest Engineering from the University of Córdoba (Spain, 2008). He is currently in the last year of his Ph.D. scholarship in the Institute for Sustainable Agriculture (National Spanish Research Council-CSIC) in Córdoba. He has works in the Remote Sensing for Precision Agriculture and Weed Science (imaping) group.

His current research interests include weed science, precision agriculture, remote sensing, and agricultural application of unmanned aerial vehicles. He has been improving his expertise in image analysis in the Interfaculty Department of Geoinformatics –University Salzburg (Austria) from August to November 2016. 

About the author

César Hervás-Martínez was born in Cuenca (Spain). He received the B.S. degree in Statistics and Operations Research from the Universidad Complutense de Madrid (Spain, 1978), and the Ph.D. degree in Mathematics from the University of Seville (Spain, 1986).

He is currently the Head of the AYRNA group (Learning and Artificial Neural Networks) a Professor of Computer Science and Artificial Intelligence with the Department of Computer Science and Numerical Analysis, University of Córdoba (Spain), and an Associate Professor with the Department of Mathematics and Engineering, Loyola University Andalucía (Spain).

His current research interests include neural networks, evolutionary computation, and the modeling of natural systems. 

About the author

Francisca López-Granados was born in Córdoba (Spain). She received the B.S degree in Biology and Biochemistry (1985) and the Ph.D. degree in Agronomy (2001) both from the University of Córdoba (Spain). She developed a post-doctoral stay at Rothmasted Experimental Station (United Kingdom, 1994-1995) working in persistence and secondary dormancy of volunteer oilseed rape seed. She is currently a research scientist at the Institute for Sustainable Agriculture (CSIC-Córdoba, Spain) and Head of the imaPing Group (Remote Sensing for Precision Agriculture, www.ias.csic.es/imaping).

Her main research lines are centered on modeling the cropping systems and the optimization of agrochemical applications by using site-specific strategies and different methodological and technological applications (remote sensing, machine learning, geospatial analysis). 

Journal Reference

María Pérez-Ortiz 1,3 , José Manuel Peña1 , Pedro Antonio Gutiérrez2 , Jorge Torres-Sánchez1 , César Hervás-Martínez2 , Francisca López-Granados1 . Selecting patterns and features for between- and within- crop-row weed mapping using UAV-imagery.  Expert Systems with Applications, Volume 47,  2016, Pages 85–94. 

Show Affiliations
  1. Institute for Sustainable Agriculture, CSIC, P.O. Box 4084, 14080-Córdoba, Spain.
  2. Department of Computer Science and Numerical Analysis, University of Córdoba, Rabanales Campus, C2 building, 14071-Córdoba, Spain.
  3. Department of Mathematics and Engineering, Universidad Loyola Andalucía, third building, 14004-Córdoba, Spain.

 

 

Go To Expert Systems with Applications

 

Check Also

A vehicle routing problem arising in unmanned aerial monitoring - Advances in Engineering

A vehicle routing problem arising in unmanned aerial monitoring