Design Of Intelligent Factory For Hydroponics Based On Internet Of Things And Machine Vision Technology

Hydroponics is a high-tech,high-yield,high land space utilization of new plant soilless cultivation technology.Based on the Internet,the intelligent plant of hydroponics is used to realize the automatic monitoring and control of plant planting environment and the prediction of plant growth maturity by using modern sensor detection technology,network communication technology,CCD computer vision technology and image processing algorithm.Hydroponics cultivation using multi-layer cultivation rack,the use of LED fill light,according to different plants in different periods of time required to develop a reasonable light intensity plan.The temperature of the plant,the concentration of carbon dioxide,the humidity of the plant,the intensity of the plant cultivation area,and the oxygen content of the nutrient solution required for plant growth,the PH and concentration of the nutrient solution will have a greater impact on the plant growth,Through the sensor detection technology to perceive the environmental parameters of the plant plant,combined with Internet of things communication technology to collect the environmental parameters pushed to the remote server,and then pushed by the server to the PC browser and mobile client to achieve plant plant environment parameters Remote real-time monitoring.The system has the characteristics of plant image acquisition and wireless video transmission,and uses the image processing algorithm to obtain the characteristics of plant leaf projection coverage area and leaf color.Then,according to the characteristics of image processing,the machine learning algorithm is used to predict the maturity of plant growth.In this paper,the following research work is carried out on the aspects of image processing algorithm,plant growth maturity prediction,such as monitoring and control of environmental parameters of aquatic plantation intelligent plant,plant bladeprojection area,color and so on.(1)Design and build the Internet-based remote monitoring system,including the ARM controller and sensor network composed of environmental parameters to monitor and control the embedded hardware and software systems,based on the Linux server to build,subscription model MQTT Internet server,while In the Linux server to build for the Internet of things monitoring services TOMCAT WEB server and MYSQL relational database,to the plant within the parameters of the automatic detection of information and real-time remote monitoring purposes.(2)According to the characteristics of area and color in the process of plant growth,the difference of plant target extraction based on image grayscale,image edge detection and color image clustering is studied.The machine learning algorithm based on k mean clustering divides the similar pixels in the image into several categories and extracts the categories of plant leaves from them.Finally,the projection coverage area and color of plant leaves were calculated based on the extracted plant target images,and the repetitive errors of image measurement under different ambient light conditions were verified.(3)Using the image processing algorithm to calculate the projected area,color and other characteristic data of the plant leaves,the artificial method was used to determine whether the plants under different characteristics were mature,and a certain number of samples were collected.Plant growth maturity prediction Machine learning training data set,The logistic regression model and the neural network model were used to train and test the training set samples respectively.The results show that the success rate of plant growth maturity is up to 99.0%by means of neural network model,which can meet the design requirements of the system.