Assessing the state of maturation of the pineapple in its perolera variety using computer vision techniques

Computer vision systems allow identifying physical characteristics and product defects in a non-invasive and reliable form. Due to these advantages, computer vision systems have been widely accepted in the agricultural and food industries, since these industries require a high demand for objectivity, consistency and efficiency in the quality control of the product, requirements that can be met by the computer vision systems. This paper proposes a method for automatically evaluate the state of maturation of the perolera variety pineapple (Ananas Comosus) in post-harvest using computer vision techniques. The proposed evaluation procedure is implemented through a digital color-image processing based on the stages of preprocessing, segmentation, feature extraction and statistical classification. For this purpose we use images in the HSV color space, segmentation by automatic thresholding using Otsu's method, the first-order moment of the distributions of the H and S planes as features, and the Modified Basic Sequential Algorithmic Scheme (MBSAS). 1320 images were utilized, which 770 images were used in the process of training and 550images in the evaluation process. The results of the evaluation procedure proposed in this paper were compared with the value judgment of three experts, showing that this algorithm has efficiency in the assessment close to 96.36%.