Using techniques of data mining for detecting water-stressed and disease-infected potato crops

点击翻译按钮获取中文摘要

Mission Geospatial (MGeo) is a company with extensive experience in providing professional and technicalconsulting services in different domains of surveying, remote sensing, and geospatial information systems(GIS). They are seeking a feasible, affordable and sustainable solution for monitoring the health status ofpotato crops. Therefore, MGeo has contacted Dr. Shahbazi to conduct a research project; the aim of this projectis using techniques of data mining for detecting water-stressed and disease-infected crops. The input dataincludes imagery captured at multiple spectral bands and three-dimensional (3D) models reconstructed fromimages. The advantages of this solution are fourfold: i) it does not attempt to establish a direct mathematicalrelationship between spectral information and biophysical attributes of plants; ii) the characteristics of stressedor diseased plants and the soil in which they are growing are learnt indirectly using data-mining techniques; iii)the effect of soil features on the plants is acknowledged during this learning process; and iv) by utilizing 3Dinformation, the impact of canopy structure and height on the results is considered as well. The developedsolution for MGeo will meet their clients' needs in terms of efficiency (applicable in large industrial scales),affordability (accessible to both large industry groups and smallholder farmers) and precision (reliable andrepeatable with known accuracy limits). Also, this solution for automatic crop disease detection offers apromising step towards sustainable agriculture which promotes both economic stability for Canadian farmersand food security for Canadian citizens.