SARS software for analysis of radiographic images of Urochloa decumbens (Stapf) RD Webster seeds

Abstract

Abstract: Recently is growing the need for non-invasive, fast, and accurate technologies that can predict seed quality. Between these technologies, X-ray image analysis stand out for evaluation of the internal morphology of the seeds. Thus, the aim of the present study was to evaluate the efficiency of a specialized software for analyzing digital radiographs of Urochloa decumbens seeds called SARS (Sistema de Análise de Radiografias de Sementes - Seed Radiograph Analysis System). Five comercial seed lots of U. decumbens cv. Basilisk were used. The seed lots were produced in the 2018/2019 crop season. Radiographic images of the seeds were analyzed in SARS, through which physical characteristics were obtained. The seeds were then subjected to germination test, in which variables related to the physiological quality were evaluated. It was possible to observe that the seeds with greater germination and vigor showed strong and significant correlations with some of the physical variables obtained using SARS. Thus, high correlation of seedling length and relative seed density is important for validating the seed radiographic image analysis method. SARS proved to be an efficient tool for analyzing digital radiographs of U. decumbens seeds. It can generate descriptors which support morphometric and internal analysis of the seeds. Physical parameters obtained by using the technique have close relationship with the germination and vigor of the seeds.