Non-Invasive measurement of signals transmitted by Plants using Kalman neuro Texture Analysis

在植物工厂等受保护植物生产系统中，控制应用仅限于环境控制。开发并成功实现了温室温度、湿度、辐射强度、二氧化碳浓度等环境因子的反馈控制技术。利用这项技术，可以通过调整环境因素来优化或控制植物的生长。植物通常会对环境参数的变化做出反应。例如，气孔活动对环境湿度和CO_2浓度敏感。植物组织的刚性受根区水分有效性的影响，应根据植物对环境的反应来控制环境因素。生物反应反馈控制系统的开发一直是植物生产工程师和科学家面临的一项具有挑战性的任务。另一个重要方面是开发非侵入性获取生长…信息的技术更多的NG植物。在植物中使用非侵入性测量的实践对于离心机人工气候室的生物响应反馈和/或前馈控制系统是必不可少的。例如，目前还没有用于直接测量植物水分状况的非侵入性方法。一种替代方法是使用间接测量技术。那些擅长种植植物的人可以在植物枯萎之前，通过植物外观的微小变化，通过它们的颜色和色调来感知他们的植物是否处于足够的水分条件下。从植物的外观来预测叶的水势是可能的。植物冠层或植物群落的外观因生长而发生的变化反映了植物群落的色调变化。音调变化可以以检索的形式以电子方式转换为图形信息。一些图像特征可以与植物冠层的色调特征相关，该色调特征也基本上反映了植物的生长状态。机器视觉可以用来监控植物的生长。连续捕捉植物在其生命周期中的图像，可以监测并可能及早发现缺陷。基于机器视觉获得的植物图像特征来评价植物的生长和健康状况，需要一种能够识别植物生长阶段和诊断胁迫症状的智能信息处理系统，为了实现开发这样的生物响应反馈控制系统的目的，首要任务是开发一种有效和实用的图像特征提取技术，最终指示植物的生长状态。在本研究中，对视网膜进行了纹理分析和有限元分析。测试结果表明，纹理分析中存在的问题是构造共生矩阵的灵活性太大，共生矩阵的构建需要极长的计算时间，提取的有限元特征清晰地反映了植物生长对图像纹理的影响。在分辨率和处理时间方面，有限元视网膜的性能优于纹理分析方案。较少

In a protected plant production system such as a plant factory, the control applications have been limited to environmental controls. The feedback control technology for greenhouse environmental factors such as temperature, humidity, radiation intensity, carbon dioxide concentration and so forth has been developed and successfully implemented. Using the technology, plant growth can be optimized or controlled by adjusting environmental factors. Plants normally respond to change of environmental parameters. For example, stomata activity is sensitive to ambient humidity and CO_2 concentration. Plant tissue rigidity is affected by the availability of water in the root zone.Environmental factors should be controlled based upon the response of plants to them. The development of bio-response feedback control system has been a challenging task for plant production engineers and scientists. Another important aspect is the development of non-invasive technology for acquiring information of growi … More ng plants. The practice of using non-invasive measurements in the plants is essential for the bio-response feedback and/or feed-forward control system of a centrifuge phytotron. No non-invasive method for direct measurement of plant water status, for instance, is currently available. One alternative uses an indirect measurement technique. Those who are skilled in growing plants can sense whether their plants are under adequate water conditions or not, from minor changes in the appearance of their plants through their color and tone, before the plants wilt. It may be possible to predict the leaf water potential from the appearance of plants.Changes in appearance of a plant canopy or a community of plants due to the growth reflect tonal variations over the community of plants. The tonal variation can be transformed into pictorial information electronically in retrieval form. Some image features can be related to the tonal characteristics of the plant canopy that also substantially reflect the plant growth status. Machine vision can be used to monitor plants growth. Continuous capturing images of plants during their life cycle allows monitoring and possibly early detection of defects. To evaluate the growth including the health of plants based on image features of plants obtained by machine vision system, an intelligent information processing system is required that will be able to identify the plant growth stage and diagnose symptoms of stress.In order to achieve the aim of developing such a bio-response feedback control system, the primary concern should be to develop an effective and practical technique for extracting image features which can eventually indicate the plant growth status. In this study, the textural analysis and the finite element retina were tested. The test results showed that problems in implementing the textural analysis are that there is too much flexibility to construct the co-occurrence matrix and the construction of the co-occurrence matrix requires extremely long calculation time.The result also showed that extracted finite element features clearly indicate the change in texture of the captured image due to plant growth. The performance of the finite element retina was better than the textural analysis scheme in terms of resolution and processing time. Less