Abstract:

In view of the ecological observation needs of Inner Mongolia Autonomous Region, especially the grassland ecosystem observation needs, forage observation was selected as the research content, and the idea of real-time remote controllable and visual forage observation automation technology based on embedded technology and parallel computing technology were proposed. The characteristic pixel image recognition method was used to realize the automatic measurement of forage coverage and the automatic recognition of forage growth image. The image height to object height transformation algorithm was obtained by using photogrammetry technology and image processing technology to realize the automatic measurement of layer height and plant height. Combined with modern electronic technology, DSP technology and image processing technology, an intelligent sensor for forage automatic observation was designed, which completed image acquisition, processing and pattern recognition at the observation front end, and then transmitted the processed results to the server through the network. Users could access the server through smartphone or PC to view the observation data remotely. The system realized the automatic observation of forage coverage and layer (plant) height and forage growth period. The test results show that the image processing is significantly correlated with the measured coverage value, with the test R_0.05>0.90, and significantly correlated with the measured plant height, with the test R_0.05>0.85. Compared with the image processing in the back-end PC system, the intelligent sensor for forage automatic observation can not only improve the data operation and processing speed of the observation system, but also have good robustness. At the same time, the intelligent sensor for forage automatic observation can greatly reduce the pressure of data transmission, and is suitable for tasks with high requirements of volume, power consumption, working environment and stability.

Key words: forage observation, embedded technology, parallel computing, forage coverage, plant height, image processing, pattern recognition