Design And Realization Of Vegetable Seed Quality Inspection System Based On Machine Vision

Seed is the most basic means of agricultural production,its importance to China is self-evident.The development of agriculture and economy is directly influenced by quality of seed,especially in China,which has a big population in the world.With the rapid development of economy,people require the quality of food nowadays other than just feeding on it as past days.However,in recent years,investigations show that China is still in a passive position in terms of competition in the market share of vegetable crops.The quality of seed is the key to improving Chinese agriculture and its competitiveness in the world,and in case of that,the selection of seed becomes a top priority for agricultural production.Because of the restrictions of the economic and scientific level,the traditional seed quality testing has a low efficiency,big harm of seeds,certain harm to detect personnel,and is process trival,uneasy to operate.With the development of automation,intelligent,digital,traditional screening methods have been unable to meet the needs of agricultural production,so we can be based on advanced technology to explore the implementation of new quality detection method.Represented by this article uses the cucumber seeds as a sample,exploring seed quality based on visualization of mechanical technology,intelligent digital filtering technology,design the corresponding detection system,and to evaluate the system.First of all,the technology based on machine vision to the vegetable seed extract characteristic values of various integrate the status of the research and analysis,combining with the advantages and disadvantages of different ways of extraction,according to the known to design a preliminary feature extraction design scheme.Secondly,by the seed for image processing to achieve the requirement of the needed for feature extraction.According to the study of different image processing methods,selecting suitable gray treatment,image filtering,edge detection,image segmentation and the other different treatments.According to the design of the eigenvalue extractiondeal with good image,extraction of ideas for the area,perimeter,elongation,roundness,color and the other several aspects of characteristic value and establish the database,and use to choose a good sample seeds and experiments verify the feasibility of this method.Then this article chooses the most widely used BP neural network system in artificial neural network.At the same time,in order to enhance thetraining effect,this paper adds the additional momentum method and the adaptive adjustment of the learning rate in the BP neural network before training the sample data.Finally,through the visual development tool system to design seeds eigenvalue extraction system based on the platform of seed eigenvalue extraction,screening of various module as system construction,then design and operation processes.Through this platform to get the purpose of displaying the visual of vegetable seed eigenvalue and classification.Experimental results show that,this article is based on mechanical visualization technology to identify the seeds of sorting system can fast accurate identification of seed feature extracting,separating showed satisfactory effects.