Studies Of Greenhouse Melon Friut Morphology Modelling Based On Computer Vision Technology

Crop growth simulation modeling is an emerging research field with the intersection of crop research science and computer science. And first-hand significant observation data of crop growth process which directly affect the model's accuracy and availability is required by the establishment of crop models. Aim to avoid the disadvantage of long data acquisition cycle is long, heavy workload, and large errors existing in current greenhouse muskmelon simulation modeling, and with the widely use of crop information extraction technology which based on the of computer version technology and the advantage of its high precision, low consumption of human resources and less affection by weather and other climate factors, this study use the own-developed image acquisition device to capture all-weather, high-precision continuous image of greenhouse muskmelon fruit, and then proceeds characteristic information of fruit morphological changes from the fruits images. At the same time, combine with the temperature and irradiation accumulated during fruit growth, based on the quantitative analysis of fruit morphological changes and the use of system analysis and mathematical modeling techniques to study the quantitative description method of greenhouse muskmelon and then to establish the dynamic simulation models based on temperature and irradiation. The main results are as follows:1. Use greenhouse muskmelon as research materials, obtain a continuous recording digital images of greenhouse muskmelon growth cycle by the own-developed acquisition device, and use MATLAB software and data processing software such as Microsoft Excel process those images and abstract the vertical diameter, horizontal diameter and fresh weight of muskmelon fruit, and finally establish dynamic growth model about greenhouse muskmelon vertical diameter, horizontal diameter and fresh weight with accumulated total production of thermal effectiveness and PAR (TEP) as model driven. And also, and also tested and compared the modeling effect of muskmelon fruit size between the TEP simulate method and the growing degree days (GDD) method through independent testing data. The results show that the root mean squared error (RMSE) between the simulated and measured results for the vertical diameter, horizontal diameter and fresh weight of greenhouse muskmelon, respectively, was 3.5273mm, 4.6959mm, and 35.5242g. Compared to the simulation based on GDD, the prediction accuracy of the vertical diameter, horizontal diameter and fresh weight of greenhouse muskmelon with our model increased by 34.98%, 21.25% and 28.34%.2. Use greenhouse muskmelon as research materials, obtained a continuous recording digital images of greenhouse muskmelon growth cycle by the own-developed acquisition device, and use MatLab software process those images and abstract the color value of muskmelon fruit superficial, and two color space RGB and HSI were used to describe the color value. Finally, dynamic change model about greenhouse muskmelon superficial color value with TEP as model driven was established. And also tested and compared the modeling effect between the TEP simulate method and the GDD method through independent testing data. The results show that the RMSE between the simulated and measured results for R value, G value, B value, Saturation (S) and Intensity (I) respectively, was 0.9990,2.6899å'Œ2.9924, 0.033,5.51. Moreover, compared to the simulation based on GDD, the prediction accuracy of the each color value with our model increased a lot. 3. Use greenhouse muskmelon as research materials, obtained a continuous recording digital images of greenhouse muskmelon growth cycle by the own-developed acquisition device, and use MatLab software process those images and abstract the texture characteristic value of muskmelon fruit with gray level co-occurrence matrix (GLCM) , and four angular second moment (ASM), entropy (ENT), contrast (CON) and coverage rate (CR) were used to describe the texture characteristic value of muskmelon fruit. Finally, dynamic change model about greenhouse muskmelon texture characteristic value with TEP as model driven was established. And also tested and compared the modeling effect between the TEP simulate method and the GDD method through independent testing data. The results show that the RMSE between the simulated and measured results for ASM, CON and CR were 0.0773, 0.063 and 0.0146, respectively.