ABSTRACT
We propose a segmentation algorithm for raisin extraction. The proposed approach consists of the following aspects. Deep learning is used to predict the number of raisins in each connected region, and the shape features such as the roundness, area, X-axis value for the centroid, Y-axis value for the centroid, axis length and perimeter of each region will be used to establish the prediction model. Morphological analysis, based on edge parameters including the polar axis, polar angle and angular velocity, is applied to search for the suitable break points that are useful for identifying the dividing lines between two adjacent raisins. To make our segmentation more accurate, some machine-learning algorithms such as the random forest (RF), support vector machine (SVM) and deep learning (deep neural network, DNN) are applied to predict the number of raisins and to decide whether the raisins need more segmentation. The performance of the three models is compared, and the DNN is the most accurate.
KEYWORDS
raisin extraction; segmentation algorithm; deep learning; image analysis; food quality grading
