In this study, a 400-1000nm hyperspectral camera can be used, and the products of Hangzhou Color Spectrum Technology Co., LTD
FS13 conducts related research. The spectral range is 400-1000nm, and the wavelength resolution is better than 2.5nm, up to 1200
Two spectral channels. Acquisition speed up to 128FPS in the full spectrum, up to 3300Hz after band selection (multi-zone support
Domain band selection).
The quality of beef is usually graded according to its marbling, physiological ripeness, muscle color and fat color. Among them, the richness of marbling is the most important evaluation index. With the development of computer and image processing technology, machine vision has been widely studied and applied in beef marbling image segmentation, feature extraction and automatic grade determination. In the automatic determination of beef marbling grade by using machine vision technology, accurate segmentation of marbling is the basis of automatic determination of beef grade. Although a large number of beef image algorithms have been proposed, the problem of inaccurate segmentation of marbling still exists due to the light source, water reflection on the sample surface and other reasons. At present, hyperspectral image technology has been used in many research fields, and there have been extensive research reports in the detection of agricultural products, but the research of beef marbling segmentation by hyperspectral image and spectral information technology has not been reported. In this paper, by mining and analyzing the spectral information of fat and muscle in beef hyperspectral images, the optimal band image suitable for beef marbling segmentation was extracted, and then the marbling segmentation was carried out by Otsu automatic threshold method, and the segmentation accuracy was compared and analyzed.
In different bands, the spectral reflection intensity of fat region and muscle region in beef hyperspectral images is obviously different. The ratio of spectral reflection intensity between fat region and muscle region is mainly concentrated in the band of 465 ~ 580 nm. At the wavelength of 534 nm, the ratio of spectral reflection intensity between the fat region and the muscle region has the maximum value. The same image processing method was used for marbling segmentation of beef color original image and 534nm feature wavelength image. The marbling segmentation accuracy η=0.928 for feature wavelength image is closer to 1 than the original image, and the mean square error is also smaller. Therefore, compared with the original image, marbling segmentation of beef characteristic wavelength image can obtain higher segmentation accuracy.