Ajuste polinomial para refinamento do resultado de segmentação de imagem para cálculo de volume de ovos

Abstract

The poultry industries and egg-related biological research are constantly facing the demand for new precision techniques to obtain egg attributes such as volume. These attributes are acquired through image processing and calculations derived from them, seeking to minimize intrusion in the process. As large amounts of image processing and calculations are performed, it is common for errors and inaccuracies to occur when processing and segmenting images containing objects, which affects the accurate acquisition of egg data.

In order to present new insights to mitigate these problems, this work addresses the application of curve fitting in the calculation of egg volume using its noisy edge points. By investigating how curve fitting can improve the accuracy of these calculations, it is possible to reduce the impact of noise on the obtained points. For this purpose, polynomial functions of different degrees are used to obtain the best fit for the set of points. Thus, with the use of a dataset obtained from several eggs, and with their volumes calculated by the water displacement method, we obtain the curve fitting performance comparable with the volume calculated from the unadjusted points, which will result in an egg shape closer to the real one, and therefore obtaining a reliable tool for volume calculation accuracy to improve future research on egg quality control.