Segmentation method of corn germ in hyperspectral reflectance image based on band ratio method

Abstract

The invention discloses a band ratio method based maize embryo segmentation method in a high-spectral reflection image. The high-spectral image of a maize seed is collected with the embryo surface of the seed upwards; multiple pixels and spectrums thereof are extracted from the endosperm and embryo and normalized; the spectral ratio of two random wave bands is calculated and serves as the band ratio, and the optimal band ratio is selected; a band subset is selected, variance of all pixels in the band subset is calculated, a seed binary image is obtained via threshold segmentation, and a background removed high-spectral image of the maize seed is obtained via mask operation; and a gray scale image of the background removed high-spectral image of the maize seed under the optimal band ratio is calculated via the high-spectral image, the gray scale image is segmented in a region growing method to obtain an embryo binary image, the embryo binary image is processed to obtain a seed embryo binary image, and the high-spectral image of the maize seed embryo is obtained via mask operation. The method can be used to segment the maize seed embryo in the visible near-infrared high-spectral image effectively, the segmentation effect is good, and the practicality is higher.

Description

technical field [0001] The invention relates to a method for segmenting corn embryos in a hyperspectral reflection image based on a band ratio method, which can be applied to the quality detection of various yellow corn seed embryos, especially to the fields of chemical characteristic analysis of embryos and the like. Background technique [0002] The embryo of the seed is the most important part of the seed, which can develop into the root, stem and leaves of the plant. If the embryo of the seed is damaged, it will have the greatest impact on the subsequent development of the plant. The embryo of corn seeds is large, the volume accounts for about 1 / 3 of the grain, and the weight accounts for 10%-20% of the grain. Corn germ contains more hydrophilic groups and is more hygroscopic than endosperm. At the same time, 84% of the fat, 83% of the inorganic salt, 65% of the sugar and 22% of the protein in the corn germ are concentrated in the corn germ, which is rich in nutrition,...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to propose a method for segmenting corn embryos in hyperspectral reflectance images based on the band ratio method, which overcomes the need for manual selection of seeds in traditional hyperspectral image analysis. The problem of artificially extracting the spectrum of the embryo after the embryo area, to achieve accurate and fast segmentation of the corn seed embryo in the hyperspectral image

Images