Journals Proceedings

Abstract

This work proposal is for localization of suspicious noise in chick’s images by selecting regions of interest. The presence of salt and pepper noise, speckle noise in an image becomes a difficult task to delimit a lesion. Some techniques of digital processing were applied in 100 chick’s images in order to minimize the noise allowing posterior extraction of the contour. The segmentation techniques used to extract the contour was performed by means of the artificial neural network self-organizing map. The various metric evaluation were used to compare the proximity of the automatically obtained area with the manual outline of a animal husbandry, resulting in percent values of 93% to accuracy, 68% to sensitivity and 98% to positive predictive value.