Automatic Leaf Motion Analysis Using Optical Flow To Diagnose Plant Behavior In Response To Environmental Changes

Abstract: We have proposed an automatic leaf motion analysis based on Infrared (IR) photography. The proposed system allows continuous, high-resolution time-lapse imaging, independent of the present of visible light, and based on open source platform. The utilization of microcomputer as a capturing unit might increase the simplicity and flexibility. The IR camera positioned above the plant as top view projection to capture the foliage images. As the motion tracking, the Optical Flow method was adopted, implementing the Lucas-Kanade technique with Shi-Tomasi corner detection. Optical flow is one of the motion detection methods to analyze the motion between two consecutive images by differentiating the constraint. Displacement distance and direction angle were used as a parameter for evaluation with regards to circadian rhythm. As the result, the proposed automatic leaf motion analysis could perform a continuous IR time-lapse photography and analyze the lateral 2D motion using the Optical Flow method. The highest probability of displacement distance observed at 6px (48%) and the direction angle was at 180° (39.2%). According to the visual appearance of time series data, the motion displacement seems to have a regular pattern resembles the circadian rhythm with regards to environmental changes although the noise also appears. To verify the relation between displacement distance, motion angle and circadian rhythm, analysis of phase and amplitude of data using a Fast Fourier Transform Nonlinear Least Squares (FFT-NLS) method will be in our future works.

Key Words: Plant motion, optical flow, leaf motion, circadian rhythm, motion analysis

Study on Plant Motion Measurement and Analysis Utilizing Computer Vision: Development of Automatic Plant Motion Analysis using Optical Flow

Plants, like other living things, have biological clocks that allow them to respond the changes in time and environment1). At day, plants do photosynthesis to produce carbohydrate such as sugar. While at night, plants distribute the stored the carbohydrate to grow leaves, stem flowers, fruits, and roots. Analysis of plant growth and motion at night might be used as an indicator of plant response due to environmental change and conditions.The objective of this study was to develop an automatic plant motion analysis with machine vision applying optical flow technique. The night photos were captured with each 30 min. interval by using an infrared camera (Pi NoIR) installed in the Raspberry Pi B+, running the Raspbian Whezzy 3.18 and infrared LED light. This capturing device was located at 1 m above the plant. The plant motions were analyzed by using the optical flow method in the OpenCV libraries. The experiment was conducted in a tomato (Solanum lycopersicum) house at the Kyushu University. Optical flow is one of the motion detection methods to analyze the motion between two consecutive images. Assuming an object pixel at time t, the same intensity part was found from the image at a time after the motion. Differentiating the constraint gives the following equation, where are the motion vector and the time derivative. The distance and direction of motion will be used as the indicator for plant diagnosis in relation with environmental data.

Oral presentation at The Joint Conference on Environmental Engineering in Agriculture 2015. Iwate University, Morioka, Japan.


Development of Smart Cloud-based Irrigation Controller: Real-time Evapotranspiration Monitoring in Tomato Greenhouse

The smart irrigation controller is a controller that reduces water use by monitoring and utilizing site environmental information and applying the right amount of water. These controllers receive feedback from the irrigated system and adjust irrigation duration and/or frequency accordingly. There are generally two types of smart controllers: climatologically based and soil moisture-based. Climatologically based controllers are also known as evapotranspiration or ET-controller. Evapotranspiration is the process of transpiration by plants combined with evaporation that occurs from plant and soil surfaces. In order to take the advantages of both type of controllers, the objective of this study is to present the development of hybrid soil moisture sensor and climatologically based controller. As part of the development of complete smart irrigation controller, real-time hourly step evapotranspiration monitoring is required on the continuous way of control. For that reason, the objective of this study was to develop a real-time evapotranspiration monitoring in the application of smart cloud-based irrigation controller.The developed real-time evapotranspiration monitoring could estimate the hourly step reference evapotranspiration (ETo) from the environmental data. The real-time computation performed in an executable function that triggered every 10 min. by implementing time shifting. Investigation on the relation between soil moisture and ETo as a potential input of hybrid controller for the smart irrigation controller is in our future works.

Poster presentation at The Kyushu Branch Japanese Society of Agricultural Machinery and Food Engineers Conference 2015
Kagoshima University, Kagoshima, Japan.

Extended Abstract


Research on Agricultural Informatics – Utilization of ICT to Improve The Agricultural Production System in Tropical Country


Environmental conditions is important factor affects the agricultural production system. In four seasons country like Japan, weather condition during cultivation can be fully controlled by utilization of greenhouse and environmental management. Microclimate manipulator can be managed automatically based on environmental evaluation. However, in tropical country like Indonesia, utilization of automatic control on greenhouse for agricultural production system is still rare because of the operational costs and the economic aspects. Thus, here we try to utilize the Information and Communication Technology (ICT) to improve the agricultural production system on tropical agriculture by implementation of regular environmental measurement using Field Monitoring System (FMS), Real-time environmental evaluation and information processing, and provision of decision support on farming activities. We realize that the incorporation of  farmer’s knowledge and utilization of ICT would be able to support modern agriculture and also promising  the progress on towards precision agriculture.

Deployment of Agricultural Information System – Installation and Utilization of Field Monitoring System on Agricultural Production in Yogyakarta, Indonesia

Agriculture is remaining to be a strategic sector in Indonesia. Besides its growth, there are also several challenges. Among others is the optimization of the agricultural production system. Practically, Indonesian farmers have improved their own cultivation technique and senses from their long-term accumulated experiences, includes knowledge about climate, weather, soil, crop type, etc. In this study, the author introduced an informatization agricultural supporting system, called the “Agri-eye” in order to improve agricultural production systems by integrating local knowledge and utilization of Information Communication Technology (ICT). Field monitoring system (FMS) is an automatic weather station installed in agriculture field to gather environmental data and forward it to the database server periodically, in addition, farmers can check their field environmental condition in real-time from a mobile phone or computer. Up to now, two FMS units have been installed in Sardonoharjo, Ngaglik, Sleman and Faculty of Agricultural Technology, Universitas Gadjah Mada, Yogyakarta as deployment test. In this article, installation and utilization of Field Monitoring System (FMS) on agricultural production will be explained. As the result, long-term environmental monitoring data (air temperature, air humidity and solar radiation) have been used as farmer decision support to prevent crops damages caused by extreme weather during cultivation. In the future works, comprehensive utilization of environmental information, local weather forecast, and control will be studied.

Keyword: precision agriculture, precision farming, tropical agriculture, information technology in agriculture, environmental monitoring

Submitted to International Conference on International Conference on Sustainable Innovation (ICoSI) Indonesia

Development of Real-time Change Point Analysis for Field Environmental Information in Agriculture

Prehension and evaluation of field environmental conditions are very important tasks in agriculture. To enhance the agricultural production system, an agricultural information supporting system ”Agri-eye” has been developed based on a cloud computing scheme. It consists of work recording, actuation and field environmental monitoring system. Regarding the environmental observation, monitoring nodes have been installed in several greenhouses in Fukuoka, Japan up to now. The environmental data has been stored in the database system in accordance. Nevertheless, the data stored in the database gradually expanded due to long-term observation. The authors have tried to utilize the feature value hidden in spacious environmental data for the present dynamical analysis. Ide and Inoue (2005) have proposed the singular spectrum transformation (SST), which uses the eigenvalues (λ) and eigenvectors (U) for a characteristic matrix formed from a set of time series data and then verified that the SST is capable of detecting the change points from arbitrary time series data automatically and systematically. Okayasu et al. (2012) applied this method to evaluate the environmental dynamics for CO2 concentration in a tomato greenhouse.. In order to dissemination of environmental information and analysis result, in this study a change point analysis using SST mentioned above was adopted in a web-based application for detecting change points from the environmental time series data. The validity of the method and performance of web-application will be verified by using the environmental data stored in the Agri-eye database measured in the real greenhouse cultivation.

Keywords: change point analysis, singular spectrum transform (SST), dynamical analysis, web application, time series analysis

Submitted to International Symposium on Agricultural and Biosystem Engineering (ISABE) 2013