| Machine vision detection technology has great potential in some industrials such as industrial design,pharmaceutical engineering due to its non-destructive and high efficiency characteristics.With the scientific and technological progress,the requirements of machine vision biomass detection technology for measurement accuracy,measurement precision and measurement stability are also increasing.Therefore,what is an urgent problem to be solved is how to improve the measurement performance of machine vision biomass detection technology to continuously meet the needs of biomass detection industry.The research in this thesis aims to improve the measurement performance of biomass morphology detection system,which is based on the analysis of the current research on the machine vision of biomass detection technology,machine technology and CAN bus detection technologyThe design of infrared sensor mechanism fusing machine vision and CAN bus is used to take real-time images of single biomass seed(corn)on the conveyor belt.The accuracy of corn seed picture plays a critical role in the accuracy of measurement system.In this paper,we studied the traditional small biomass seed size detection system,and find that there is a serious problem affecting the accuracy of system image acquisition:when the sensor gets the signal of seed particle in real time,the signal will inevitably make mistakes,resulting in the image not being acquired in real-time,affecting the efficiency of the whole system.In order to solve this problem,this paper proposes a design method of fusion machine vision and CAN bus,which has achieved optimization and improvement based on the traditional scheme and CAN bus arbitration technology.Trigger snapshot is adopted to transmit the sensor signal through CAN bus to ensure that the sensor signal will not be wrongly processed,and at the same time,the signal CAN will be transmitted to the computer accurately,so as to control the camera to obtain pictures in real time.The success rate of real-time image shooting was 86.6%.Multi-corn grain boundary line extraction algorithm based on regional growth tracking:Traditional tracking algorithm can only obtain a single target in the image,which is inefficient for a large number of corn seeds.Aiming at this defect,this paper studies and compares the existing image processing methods,and proposes an multi-corn grain boundary line extraction algorithm which is based on regional growth tracking.Experimental results show that this method can effectively reduce the image noise and simultaneously track,count and mark the boundaries of several corn seeds on a single image.A com seeds shape grading algorithm based on composite geometric features:In order to distinguish the morphology of biomass seeds(com),it is necessary to study the methods that can effectively improve the morphology grading of corn seeds.In this paper,a biomass seeds(corn)morphological grading algorithm based on composite geometric features was proposed.According to the three important parameters of seed grain area,perimeter and complexity,corn seed grain was divided into undamaged and damaged.Three characteristic parameters of corn seed were calculated and the qualified range of characteristic parameters was established.The results showed that the discrimination accuracy of intact com and damaged com was 93%and 95%respectively.With the help of platform VS2008,the interface of the design of infrared sensor mechanism fusing machine vision and CAN bus,the interface of automatic detection of com seed grain,and the interface of integrating stepping motor and eliminating mechanism of CAN bus were designed.The system can obtain the image of com seeds in real time,identify and grade the image,and eliminate it,which lays a technical foundation for the batch grading of com seeds.At the same time,the system can also be used for reference for the machine vision detection of biomass seed grains such as peanut,rice and soybean. |
PDF Full Text Request