Study On Parameters Of Maize Breeding Traits Based On Image

Characters of corn ear and grain is an important reference index for maize breeding, Traditional character parameter acquisition is mainly dependent on manual measurement and calculation,which produces many subjective errors and low efficiency. Based on the image of maize breeding parameters acquisition research combined with machine vision technology and image processing technology to extract the characteristics parameters of maize, which is the basis for the realization of the automation of corn test. Therefore, it is of great practical significance to study the research of automatic, digital and high efficiency for the realization of corn character parameters by using image.On the basis of investigation and analysis on the research status and actual demand of the high efficiency automatic test. This paper puts forward a method for obtaining maize traits parameters by processing corn ear images and maize grain image. The results obtained in this paper are as follows:1) Design a corn ear image acquisition system and the corn kernel image acquisition system. The system combines the actual requirements of corn ear and kernel image feature extraction, to develop the best image preprocessing scheme. Including image grayscale, image enhancement, edge detection and image segmentation.2) Propose an improved watershed segmentation algorithm based on the distance transform of the two valued image and the local minimum of the merged image,this algorithm can quickly and accurately segment the large accumulation of corn grain.3) Propose a fast estimation of row number and ear row number model, based on the biological characteristics and color characteristics of the ear and combined with contour tracking algorithm, to realize the accurate calculation of the corn ear row number and ear row number.In this paper, the method of corn ear and kernel image preprocessing and the ear and the grain character parameter extraction method is verified by the experiment. The experimental results show that the image preprocessing results meet the image feature extraction of image, the improved watershed algorithm is proposed to achieve accurate segmentation of adhesion of the corn in the image, the successful segmentation rate was 97.6%. At the same time, The zero error rate between the measured value of corn ear row number and grain number per row and the true value controls in 96.8%. Therefore, The segmentation algorithm proposed in this paper has certain reference significance to the research of image segmentation. The measurement method of the corn ear row number and ear grain number provides an effective reference for study on yield component traits of maize.