Real Time Specific Weed Classifier Based on Variation in Environmental Conditions

ABDULRAHMAN, ALHARBY and ATIQ UR, RAHMAN and IRSHAD, AHMAD (2015) Real Time Specific Weed Classifier Based on Variation in Environmental Conditions. In: Third International Conference on Advances in Computing, Electronics and Communication - ACEC 2015, 10-11 October, 2015, Zurich, Switzerland.

20151023_094309.pdf - Published Version

Download (780kB) | Preview
Official URL:


Natural light is an important factor that needs to be considered in the implementation of an automatic sprayer system. This paper evaluates the scalability of real time specific weed recognition system using histogram analysis, and its comparison with two dimensional weed coverage rate (2DWCR), and Angular Cross Sectional Intensities (ACSI) classifiers based on the variation in the natural lighting conditions to encompass the cloudy and bright shiny outdoor environment. A large image dataset of 1500 images was used as compared to the previously used image datasets (200-1200 image datasets). The dataset images are further subdivided based on different lighting conditions and is termed as normal, dark, and bright. The proposed classifier was applied to classify these images into broad and narrow class for real time selective herbicide application using a single constant threshold. The analysis of the results shows over 94 percent weeds detection and classification accuracy (broad and narrow). The results confirmed the scalability of the proposed classifier to encompass dark cloudy as well as bright light outdoor conditions on the same day of application, while maintaining the same accuracy and 22% - 30% better classification accuracy than Two-Dimensional Weed Coverage Rate and Angular Cross Sectional Intensities classifiers.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Image Processing, automatic sprayer system, weed classification, histogram analysis
Depositing User: Mr. John Steve
Date Deposited: 20 Apr 2019 11:36
Last Modified: 20 Apr 2019 11:36

Actions (login required)

View Item View Item