A Practical Approach for Segmentation of Twigs on Samples of Yerba Mate by Digital Image Processing

This paper discusses the implementation of the digital image processing techniques as a method for estimating the contents of twigs in samples of yerba mate. The hardware used in this paper, to obtain the analyzed images, consists of a desktop scanner and a personal computer. Then, using the designed and developed software for this purpose, the digital images obtained were processed and analyzed using a basic segmentation technique such as binarization with histogram thresholding. This paper compares two methods for estimating weight ratios from the binarized image. Preliminary experimental results using the proposed techniques, suggest that it may be estimated with an error less than 4% the percentage of twigs in a given batch of images. The speediness, simpleness and efficiency of the test, compared with current techniques, as well as, the reduced cost of the associated hardware, presents a viable, inexpensive and straightforward alternative if compared with the traditional.