Sistema de análise de sementes (SAS) na detecção de misturas varietais e de sementes esverdeadas em soja

Abstract

Soybean is a crop of great economic and social importance to Brazil. The use of high quality seeds at sowing is fundamental to obtain high productivity.The presence of green seeds and varietal mixture that occur during the production process has affected the quality of commercial lots. Image analysis through computer systems is a promising technique, non-destructive, rapid and objective that has been used to evaluate the seeds quality. Currently, it was developed by the company Tbit the SAS equipment (Seed Analysis System) that has the potential to analyze the physical and physiological seeds quality. The present work was realized with the objective to verify the possibility of using image analysis using the SAS equipment and compare their performance with traditional methods used in assessing the physical quality of soybean seeds lots. For the green seeds analysis was used samples with predetermined percentage of green seeds mixture (5% to 50%).The SAS efficiency of identify green seeds was 99.51%.In detection of varietal mixtures was used morphological description of seeds, as well as visual analysis of seed and their images. Decision networks were used based for the main differential characteristics among the cultivars. It is possible use the SAS® equipment for differentiation of soybean cultivars, especially when evaluating the color of hilum.