Fundamental research far an intelligent plant factory based on the computerized network.

本研究探讨了一种有效的数据通信技术，适用于由多个植物工厂组成的分散植物生产系统，该系统工作在网络上。图像数据具有尺寸大的特点。为了在网络上处理如此大的数据量，提出了一种有效的植物图像压缩技术。该系统通过传输不同植物工厂栽培的植物图像，用于诊断植物的生长状况。该技术对于多个工厂系统之间的数据通信非常有用。(论文1)2。提出了一种立体传感技术，在实时处理中构建三维植物形态。首先利用两台CCD相机分别从左右两个方向采集图像，合成水果图像，然后用直线显示提取轮廓。通过这些步骤，我们得到了一个三维植物模型。该技术大大缩短了三维图像构建的处理时间。利用远距存在法的概念，我们可以得到一个更有效的三维模型，它看起来像一个真实的人类视觉所拍摄的图像。（论文2、3）开发了一种用于收获机器人视觉传感器的图像识别技术，并将其应用于带式输送机上移动秧苗的质量评价。该技术为评估提供了高速处理（5张图像/秒）。该技术有助于实现工厂评价系统的自动化。(4) 4。水果的形状评价具有很大的经验性和不确定性。提出了一种利用吸引子、分形维数和神经网络定量评价果实形状的新方法。果实形状鉴定误差与吸引子形状和分形维数之间存在高度相关。因此，这些方法可以定量评估果实形状的复杂性，并有助于植物工厂之间的数据通信。(5) 5。最后，利用上述新开发的图像处理技术，对分散植物工厂的全面机械化和自动化进行了研究。（文件6）更少

In this study, an effective data communication technique for a decentralized plant production system consisting of many plant factories, which works on the network, was explored.1. Image data are characterized by large size. An effective image compression technique for plant was developed to deal well with such large sized data on the network. It was applied for diagnosing the growth conditions of plants cultivated in different plant factories by transmitting their plant images. This technique was very useful for data communication among several plant factory systems. (Paper 1)2. A stereo sensing technique was developed to construct 3-dimensional plant shape in real-time processing. A fruit image was first obtained by composing two images took from the right and left directions using two CCD cameras and then extracted its outline by displaying with linear lines. From these procedures, we obtained a 3-dimensional plant model. This technique significantly shortened the processing time fo … More r 3-D image construction. We could obtain a more effective 3-dimentional model, which looks like an image taken by a real human vision, by using the concept of a tele-existence method. (Papers 2 and 3)3. An image recognition technique for a visual sensor of a harvesting robot was developed and applied it to the evaluation of the quality of seedlings which are moving on the belt conveyor. This technique provided a high-speed processing (5 images/s) for evaluation. This technique was useful to achieve automation of a plant evaluation system. (Paper 4)4. Shape evaluation of fruit is quite empirical and uncertain. A new technique was developed to evaluate the fruit shape quantitatively using attractor, fractal dimension and neural networks. There was high correlations among identification errors, the shapes of attractors and fractal dimensions for evaluating fruit shape. So, these approaches allow the quantitative evaluation of the complexity of the fruit shape and are useful for the data communication among plant factories. (Paper 5)5. Finally, total mechanization and automation for decentralized plant factories were investigated using newly developed image processing techniques mentioned above. (Paper 6) Less