Construction And Verification Of Corn Internal Components Volume Model

At present,corn is one of the most important crops in the world.Its planting area and yield are second only to rice and wheat,ranking third.Corn is not only an important food crop,providing people with a variety of food,but also the most important raw material in animal feed,as well as the main industrial raw material for the production of alcohol and starch.The drying of high moisture corn is an important link to ensure the safe storage and circulation of corn.From the perspective of the factors affecting the drying cracks of corn,they are all related to the three-dimensional structure and internal distribution of corn kernel,including the distribution of horny endosperm,farinaceous endosperm and embryo.In order to study the effect of the three-dimensional structure and the density difference of different parts of corn kernel on the drying cracks,the volume model of corn kernel internal component was constructed to provide a method for estimating the volume of corn kernel internal component.Most of the research on the internal structure of objects is based on CT scanning technology,which is widely used in medicine.CT technology can detect the external and internal structural dimensions of the object nondestructive,and it is suitable for the measurement and analysis of the internal of the object.However,due to the expensive equipment,high cost of measurement and high requirements for operators,it can not be widely used in general measurement.In addition,the research on corn mainly focuses on the reconstruction of plant,leaf and canopy,while the researches on the internal structure of corn kernel are relatively few.In this study,how to get the actual distribution of horny endosperm,farinaceous endosperm and embryo in corn kernel was studied,which provides theoretical basis for the estimation of internal components volume and density of corn kernel.The main research work are as follow:(1)Six corn varieties,Xianyu 335,Zhengdan 958,Denghai 605,Lianchuang 808,Qiule 368 and Wansheng 68,were selected as the research samples,and the weight and volume of single kernel were measured.Different kinds of light sources were used to design the image acquisition device of corn kernel embryo-side and top big-end.The embryo-side transmission image and top big-end reflection image of corn kernel were obtained by analyzing different white light intensity.(2)A positioning profiling corn seed cutter was designed.The profiling groove and micro spring are used for positioning to ensure that the corn seeds can be cut along the embryo.The spacing of the profiling grooves is suitable to ensure that the cutting effect of each corn kernel does not interfere with each other.When cutting,it is cut from the small end to the large end of the kernel.The force is appropriate to ensure the cutting quality of single kernel.The corn kernel samples were cut along the embryo,and the image acquisition device of corn kernel longitudinal section was designed,and the reflection image of corn kernel longitudinal section was acquired.(3)A low contrast image segmentation algorithm for corn kernel was proposed.Based on the nonlinear transformation of color channel,the corn kernel image was transformed from RGB color space to Lab color space.After the binary image of corn kernel image was segmented under different color channels,the principle of overlapping region was used to realize the accurate segmentation of corn kernel internal components,and the two-dimensional distribution images of horny endosperm,farinaceous endosperm and embryo in the embryoside,top big-end and longitudinal section image of corn kernel were obtained.The segmentation effect of multi-channel overlapping region method and traditional image segmentation methods on different varieties of corn kernels was evaluated by the recall ratio,precision ratio and comprehensive index.The results show that the recall rate,precision rate and comprehensive evaluation index of the multi-channel overlapping region segmentation method are all over 98%,and the segmentation effect is better than the traditional segmentation methods,which can achieve the accurate segmentation of different corn varieties.(4)The volume model of components in corn kernel was established.Based on MATLAB,according to the distribution of horny endosperm,farinaceous endosperm and embryo in the embryo-side image and the top big-end image,the longitudinal section image was scaled to a certain scale to generate the multi-layer slice image of corn kernel,taking the two-dimensional distribution of the longitudinal section image as the starting point of three-dimensional growth,and taking the two-dimensional distribution of the embryo-side image and the top big-end image of corn kernel as the ending boundary of three-dimensional growth for virtual growth,The three-dimensional reconstruction of the inner components of corn kernel was realized by volume rendering,and the volume models of internal components were obtained.(5)The verification of corn kernel volume model was completed.Based on the volume model of corn kernel,the volume of corn kernel was calculated by image calibration technology.To compared with the actual measurement results of single kernel volume,the Grubbs test,F test and t test were carried out.There is no significant difference between the accuracy and average value of the two,which proves the accuracy and stability of the volume model of corn kernel of different varieties.