Research On Water Stress Measurement Of Plants Based On The High-throughput Infrared Thermal Image Processing

With the rapid development of computer vision technology, it has certain prospects to combine agriculture, forestry, plant physiology and crop cultivation theory and develope the plant water stress measurement experiment system based on computer technology. Human beings are facing a rapid increase in demand for food, water scarcity, climate extremes and other issues, resulting in a very serious impact on agriculture and forestry. In addition, when using traditional detection sensor to measure the temperature of the plant leaves in the field of botany, agriculture and forestry, we can not simultaneously measure a plurality of plants, can not simultaneously measure multiple plant canopy leaves, can not simultaneously measure multiple areas on the leaves, and can not visualize the data collected. Even the operation of reading and the installing for experimental data are complicated. Therefore, it is essential to innovate a device for water stress measurement. In this research, we studied on plant water stress measurement. The water stress measurement based on high-throughput infrared image processing can compensate for the shortcomings of traditional devices. Not only for solving the problem of plant growth under drought conditions but also for reducing the impact in harsh environmental factors. Evenmore, it has a high value in water conservation, environmental protection, agriculture and forestry irrigation and other practical work. Also, it has a direct contribution to human survival, long-term development of agriculture and forestry, increasing food production. This proposed method can be applied to the initial judgment of drought resistance for different plants. There is a wide prospect of practical application in the field of agroforestry and also the important contributions for innovative research in the field of forestry equipment and information.On the basis of theoretical analysis in infrared thermal image processing technology, plant water stress measurement techniques and tomato plant water stress, we constructed the plant water stress measurement experimental system based on the infrared thermal image camera and proposed the experimental protocol to this system for measuring plant water stress. We did the experiments for cultivated tomato Solamum lycopersicum (M82) and wild tomato Solanum pennellii (S. pennellii). The experiments were repeated for a large number of times by using tomato plants. We extracted temperature data of tomato canopy and analyze the condition under water stress by collecting and processing the infrared thermal images. Compared the canopy temperature data measured by the proposed approach with the canopy temperature obtained by traditional sensor, the same trend and higher accuracy are founded. It proved the correctness and feasibility of this method from a quantitative point of view. In addition, we built the plant water stress related models by the collected data and applied to the initial judgment of drought resistance. The conclusions by the proposed approach and the weighing approach are the same. It proved the correctness and feasibility of this method from a qualitative point of view. Therefore, the proposed method for plant water stress measurement based on high-throughput infrared thermal image processing is reasonable and effective. The main contributions of this paper, there are four aspects:First, we proposed the plant water stress measuring system based on infrared thermal image camera. According to the needs analysis of experimental system, we built the experimental environment, chose the tomato varieties, cultivated and transplanted samples, designed the size and number of plants, prepared for the instruments and equipment and tested the temperature distribution of the thermal image area. After the establishment of the experimental system, the traditional weight method and the plant water stress measurement experiment based on the infrared thermal images were implemented. Then we mapped the image gradation and temperature after data extraction.Second, we proposed extraction algorithm of plant canopy temperature based on high-throughput infrared thermal image processing and made improvements. According to high-throughput infrared image feature, we made the steps of image processing including image compression, threshold segmentation, background area clear, single trees plants cropping, individual plant thresholding, leaf area selection, measurement, wherein the threshold is the key of image processing. By using a variety of algorithms for the segmentation of many plants and background and individual plants, the experimental results proved that Bernsen algorithm is the best way to segment the multi-plant trees and background. For the individual plants, the effect was not significantly different by using most of the common segmentation algorithms. We also compared the visible image and infrared thermal image, analyzed the morphology of plant canopy, improved the proposed algorithm and proposed the extraction algorithm for plant canopy temperature based on the subregional model. According to the low temperature of new leaves and the high temperature of old leaves, we made the plant canopy regional breakdown and used concentric method to measure the temperature separately. In addition, the k-means clustering algorithm is applied to the old leaves the area and the weighted averaging value is calculated based on the area of old leaves and new leaves. It helps the temperature measuring range more comprehensive and accurate. The proposed image processing algorithms were implemented by ImageJ, an image processing platform, and Macro language.Third, we did the experiments for a large number of times by using tomato plants. It proved that the proposed method can be applied to initially judge the drought resistance and the method is reasonable and effective. We prcessed the thermal images to extract the temperature of plant canopy, compared the data with the air temperature and the temperature data extracted by the 4-channels thermocouple data logger. The result quantitatively proved that the proposed method is correct with higher accuracy. In addition, we established the canopy-air temperature difference model and water stress index model model for tomato plants by the collected data and applied the models to the drought resistance initial impression for different tomato varieties. The conclusion is the same to the result obtained by weight method. It qualitatively proved that the proposed plant water stress measurement method can be used to initially judge the drought resistance and the proposed method is reasonable and effective.In summary, the proposed approach for water stress measurement can be used to monitor the study of plant water stress and has good versatility and scalability. It can provide the design ideas, experience and theoretical guidance for study plants under water stress in the filed of forestry equipment and information.