Research And Implementation Of Count Method Of Rapeseed Based On FPGA Image Acceleration

Rape which is one of the main oil crops in China plays an important role in the production and processing of edible oil.One of the directions of rapeseed breeding research is to improve the oil yield of rapeseed.In this kind of research,rapeseed quantity measurement is the premise to determine the research data,which directly determines the reliability of the 1000-seed weight value as one of the basic research parameters.The traditional counting method is mainly completed by manual counting,which is time-consuming,laborious and repetitive.This paper proposes an application scheme that can calculate the number of seeds in rapeseed image.The scheme is based on the principle of digital image processing and FPGA technology,through the acquisition of rapeseed image,image algorithm for processing and recognition of counting,to solve the shortcomings of manual counting.De10-nano development board is a new development board with Inter FPGA chip produced by Youjin.It has a large number of basic logic units inside.It can realize the design,test and application of electronic circuits through verilog HDL hardware description language and DE10-Nano development board.In the field of digital image processing,FPGA can realize the pre-processing operation of image.Because of the programmability and good portability of FPGA design,the subsequent design of its electronic circuit will be flexible and has application value and prospect.The main work of this paper is as follows:1.Based on image provided by the academy of agricultural sciences’ fuel data as experimental object,the digital image processing after briefly introduced principle and FPGA technology,learning and chose this article listed several image preprocessing algorithms,the algorithm step by step through MATLAB data processing tools to implement,using morphological processing method of target image further processing,In order to eliminate noise,image preprocessing algorithm learning,selection and software verification.2.The counting principle and requirements of this scheme are introduced,and realized by MATLAB: rapeseed in the image is counted by using connected area quality.Due to the adhesion phenomenon in the image,this paper proposes an experiment to solve the adhesion phenomenon by using watershed algorithm,and finally completes the counting,and the success rate reaches 97.95%.3.A rapeseed image preprocessing system based on DE10 platform and Verilog hardware description language was designed.This system can collect image data from camera and PC.The project design of the development board includes image data input and output module,gray conversion module,median filter module and morphology processing module.The circuit design and optimization of each module are carried out by using the FPGA parallel characteristics,and the results are displayed.At the same time,the results are sent back to the PC and the upper computer is counted by MATLAB,so as to meet the final requirements.By comparison,the final counting accuracy of the images collected by the camera and PC PC reaches 94.12% and 96.10% respectively,basically realizing the expected.