A novel computer vision-based approach to automatic detection and severity assessment of crop diseases

Abstract

Accurate detection and identification of crop diseases plays an important role in effectively controlling and preventing diseases for sustainable agriculture and food security. In this work, we have developed a novel computer vision-based approach for automatically identifying crop diseases based on marker-controlled watershed segmentation, superpixel based feature analysis and classification. The experimental result demonstrates that the proposed approach can accurately detect crop diseases (i.e. Septoria and Yellow rust. Two types of most important and major wheat diseases in UK and across the world) and assess the disease severity with efficient processing speed.

Automatic Detection and Severity Assessment of Crop Diseases Using Image Pattern Recognition

“…It is very important to identify features of a certain disease, extract the most discriminative features and then build a classification model using suitable image processing and machine learning approaches. The feature extraction procedure has been briefed as follows which has been explained in detail in our previous paper [12]. …”

“…The marker-controlled watershed algorithm uses operations of mathematical morphology to place foreground markers in the blob and background markers for the areas without blobs. The details have been summarized in [12] but a block diagram representing the leaf extraction from a complicated background has been shown in Fig. 2.…”

“…In our case, offset value is set to 1. The features, which are calculated using GLCM matrix are summarised in [12]. The mean value in each direction was taken for each Haralick feature and they were calculated from both red and green channels.…”

Role of Big Data in Agriculture

“…Liangxiu Han, et al, 2015 Detection and identification of specific crop diseases to prevent and control the plant diseases efficiently.…”

Detección temprana de la Sigatoka Negra en hojas de banano mediante imágenes hiperespectrales: Un enfoque aplicando los métodos PLS-PLR y HS-BIPLOT

“…Las imágenes hiperespectrales de las hojas conforman un set de datos tridimensional con altas dimensiones espaciales y espectrales que tienen características distintivas como son: (1) alta colinealidad en las bandas adyacentes, (2) variabilidad de firmas hiperespectrales y (3) alta dimensionalidad debido a la elevada resolución espacial y espectral de los sensores hiperespectrales (Lu & Fei, 2014). Consecuentemente, es necesario aplicar metodologías de procesamiento multivariante capaces de correlacionar los perfiles hiperespectrales y las infecciones de las plantas con el fin de detectar e identificar en forma precisa las enfermedades en los cultivos para la aplicación oportuna de estrategias de control de plagas y prevención de enfermedades (Han, Haleem, & Taylor, 2015).…”