ASABE Technical Library - Abstract

Abstract. This paper put forward a method of using image technology to identify the vitality of vegetable seed. Tomato as a kind of important vegetable was used for the research. Seed vigor is an important indicator of measuring the quality, which is integrated performance of determining the activity level and behavior characteristics of seed during germination and emergence. Traditional seed vitality identification methods mainly include seedling grading method, determination of seedling growth rate and resistance to cold, frozen germination method, accelerated ageing method and conductance method, tetrazolium method and so on, but these methods are all susceptible to the influence of various factors and time-consuming. With the development of computer technology, the research about seed vitality identification based on image characteristics is more and more. In order to identify the vitality of vegetable seeds accurately and rapidly, an identification method based on shape, size and color extracted from the images of tomato was proposed in this paper. First of all, the original images of 60 tomato seeds were acquired under the LED light source. Then segmentation, noise removing and single extraction were carried on the original image to acquire tomato seed image clearly and concisely. And then geometric features like area, diameter, circular degree and color features based on the color models RGB, Lab and CMYK were extracted from the preprocessing image of pepper seed. And germination experiment was carried out to get germination rate, germination potential, germination index and the fifth day of seedling length of the tomato seeds in order to get vitality of the batch of seed. Finally, Principal component analysis (PCA) and support vector machine (SVM) were used to determine the relationship between image parameters and seed vigor. The experimental results show that area, perimeter, circular degree and color features can identify tomato seed vitality, and the correlation rate is all over 0.8, which indicated the usefulness of the proposed method. And the results showed that the seed vitality index increase with the increase of area, perimeter, and they have significant correlation.