Regresar

Texture analysis for green forage measurement using pasture images

Abstract:

This memoir consists of implementing a program capable of classifying textures extracted from image captures of different stages, using computer vision techniques. The investigation is divided into three fundamental parts: The GUI (Graphical User Interface), the texture characteristic extraction algorithms, and lastly computer-supervised learning. Its results will contribute to improving decision-making by identifying the best plot of land, so that the cattle can be well fed and have an improved milk production. To analyze the results, first and second order statistics were used, and the validation of the algorithm was done by comparing a pattern of fodder with a low percent of dry material and with image samples of plots of land marked as optimal and not optimal to form training data and obtain the desired results. With the resulting matrices it could be determined that the energy characteristic is better than the histogram and co-occurrence.