IMAGE ANALYSIS TECHNIQUE UTILIZATION TO ESTIMATE THE LENGHT AND THE WIDTH IN CORN SEEDS

PEDRO ABEL VIEIRA-JUNIOR, SILVIO MOURE CICERO, LUCIO ANDRÉ DE CASTRO JORGE, LUIZ CARLOS MIRANDA, DURVAL DOURADO-NETO, CARLOS TADEU DOS SANTOS DIAS E WALTER RODRIGUES DA SILVA

 

In spite of important for the sowing, mostly in precision drills, there is lack of more specific official patterns for the classification of corn seeds, mainly in relation to the length. Probably the lack of an official pattern is due to an absence of practical, exact and repetitive methods to measure the seeds. Samples of eight corn seed lots were submitted to retention sieve test and to a practical test used to measure seed length (longuímetro). In samples of seeds retained by the sieves and measured by the longuímetro, the length and width dimensions were measured being used the profile projector, a measuring instrument known by its precision and accuracy. The measures determined with profile projector were compared, with and among themselves, to the respective measures determined with methods traditionally utilized to evaluate the classification of corn seeds and to the measures estimated applying the SIARCS 3.0® software in images of seeds made in resolutions of 100, 150, 200 and 250 ppi, and with those seeds in different arrangements The following parameters and procedures were studied: resolution of digitalization, size of the sample in function of the required precision, proceeding of the software (method of the bounded rectangle
and of the ellipse that best represents the object) and arrangements of the seeds during the digitalization of the images (dorsal and ventral position, and disposition). In function of the adaptation of the parameters and of the studied procedures, it was concluded that the image analysis technique is viable to estimate measures of length and width in corn seeds, presenting high precision and satisfactory accuracy, as well as additional advantages in relation to the traditional methods.



Patrocinadores