Study On The Method Of Detecting Maize Ear Quality Based On Machine Vision

High-quality corn ear is to improve the yield,utilization and economic benefits.Currently,the use of machine vision technology for the detection of maize quality is mainly focused on the analysis of corn kernels,whereas the related research for maize ear are few.But the corn kernel can not reflect the size,shape and other genetic traits of corn ear.Therefore,this paper studied methodology for the quality analysis of maize ear based on machine vision approach.In this paper,the hardware system for collecting image data was built based on the analysis of color characteristics and the optical properties of the corn ear.Based on the characteristics of color images,an efficient preprocessing scheme has been designed,which includes selection and conversion of HSV color model,median smoothing filtering,improved entropy threshold segmentation,morphological processing,etc.Through analyzing the characteristics of corn ear,an novel approach which combines the consideration of four properties: corn size,shape,color and texture,was proposed to identify the corn ear.The boundary tracing method was adopted to extract the ear contours of maize,mark and locate multiple ear of corn.Then,the minimum circumscribed rectangle was extracted based on contour convex hull,and the length and aspect ratio of the smallest circumscribed rectangle was calculated,which was used as the characteristic parameters of maize ear size and shape.Simultaneously,the mean and standard deviation of the H component were used as the color characteristics of the ear,and the mean and standard deviation of the V component were used as the texture features of the ear.In terms of quality identification,the maize ear signature library was established,from which the data for statistical pattern recognition can be extracted.After that,the identification interval of each feature parameter was determined.Finally,the identification of normal maize ear was achieved by judging whether the feature data are in their recognition interval has thereby achieving the purpose of eliminating the exception of maize ear.Based on the powerful image processing function of OpenCV and the MFC graphical interface of Visual Studio,this paper designed and developed the quality identification and detection system for maize ear,meanwhile,the accuracy and detection rate of the system were verified.The results show that the accuracy rate can be 96.5% when the number of ear is 200,and the average detection time of each ear can up to 0.1 seconds.Therefore,the method in this paper can achieve high efficiency and accurate nondestructive testing and identification of corn ear,which has a strong practical significance and use value.