| As one of the four major oil crops, rapeseed is widely cultivated all around the world. The seed colors of rapeseed are diversiform and mainly black, brown or yellow. Compared to the black-seeded or brown-seeded rapeseed, the yellow-seeded rapeseed is characterized with thinner seed coat and lower hull percentage, higher oil and protein content, higher oil quality, lower crude fiber content. So the breeding of yellow-seeded rapeseed becomes one of the main objectives in rapeseed breeding.The seed colors of rapeseed are miscellaneous and gradually mutative from yellow to black. It is difficult to classify and distinguish the seed colors in rapeseed. Usually, the seed colors of rapeseed are distinguished, identified and classified subjectively, lacking a uniform standard to classify the seed colors automatically. In the present study, the seed colors were captured and transformed into digital information with image processing technology by computer. Combined with near-infrared spectroscopic analysis, we tried to find a way to study the seed colors of rapeseed qualitatively and quantitatively.As the near-infrared spectrometry (NIRS) analysis is superior in simple and rapid process, non-destructive for the samples, it has been widely used in today's agriculture researches. There were some researches on seeds of rapeseed by NIRS method. However, most of these studies have been limited to the following quality traits, such as oil content, protein content, fatty acids content, cellulose content and glucosinolate content. Researches on seed color with NIRS were few.In the present research,442 spectroscopy data of seeds of 306 Brassica napus samples and 136 Brassica campestris samples were collected by near infrared reflectance spectroscopy (NIRS). In the mean time, the seed colors were scanned and transformed into color digital information values (RGB forms). The NIRS mathematic models of main information value in seed colors were developed by combination of computer-aided image processing and NIRS analysis, which gave a foundation to further study the seed colors of rapeseed and provided new referrence to classify the seed color grades in rapeseed. The main results of the present study were as follows:1. In the seed color information of rapeseed(Brassica napus and Brassica campestris), R-value was the best one that could represent the diversification of rapeseed seed colors. According to eyeballing of the seed colors, the seed color were lassified into 6 grades which were grade 0, grade 1, grade 2, grade 3, grade 4 and grade 5. The corresponding R-value of seed color information in each grade varied in the range of 0-55 (grade 0), 55～100 (grade 1),100～125 (grade 2),125～140 (grade 3),140～150 (grade 4), and 150～255 (grade 5).2. Among the NIRS models of seed color information, the best one was that developed on the basis of R-value. In the predictive model of R-value, the internal cross validation test showed that 1-VR was over 0.85. In the external test of the models, determination coefficient of validation (RSQ) was over 0.83.The NIRS predictive model developed on the basis of the R-value of seed color information had high precision and stability. This model provided a basis for qualitative and quantitative analysis in seed color study and could be widely used in yellow-seeded rapeseed breeding.
|
PDF Full Text Request