Research On Path Navigation Technology Of Agricultural Machinery Based On Machine Vision

Machine vision-based navigation system has been widely used in industry,service industry and others.For the agricultural field,some existing navigation methods can not satisfy the needs of detection and navigation in complex operating environment.Therefore,the application of machine vision navigation technology and its realization in embedded system have important research significance and practical engineering value in the agricultural field.In this thesis,the agricultural transplanter as specific application is researched.A visual navigation system is designed by researching the navigation path planning algorithm.In addition,a real-time hardware system based on the DSP chip is designed to realize the algorithm in embedded hardware system.The research includes the following parts.Firstly,aiming at the problems of image information redundancy and model complexity caused by camera tilting down imaging in agricultural navigation system,a visual navigation system is designed.The visual navigation system collects scene images vertically and downwards,and based on this,a mathematical model of the visual navigation system is designed.Then,the navigation path planning algorithm is designed.In this path planning algorithm,the green feature of seedling is regarded as the target feature,and a 2G-R-B component of RGB color space is chosen for image gray processing by comparing the green segmentation effect of different color spaces.After extracting the seedling area from the image,the DBSCAN fuzzy clustering algorithm is used to obtain the seedling center point to provide parameter points for subsequent linear fitting of the least square method.Aiming at the time-consuming problem of the algorithm,the clustering algorithm is optimized to meet the real-time requirements of the system.Finally,the hardware processing system based on DSP chip is designed.DM642 is used as the core processor to complete the overall circuit system design including peripheral basic circuit module,video module,memory module and etc.With the help of DSP development tool CCS,the DSP program implementation of the algorithm is completed.The DSP implementation optimization of the algorithm is discussed systematically at the end.The algorithms studied in this thesis have been verified by experiments.The results show that the proposed algorithm has good detection effect on the basis of real-time performance.