Weight and Diameter Estimation Using Image Processing and Machine Learning Techniques on Apple Images

Abstract

Classification of table fruits according to size is traditionally hand made. But human factors are the cause of faulty classifications. Automatically performing this process with the machines is important in terms of speeding up the process, reducing costs, and minimizing errors. In this study, weight and diameter estimations were made on Starking type apples using image processing techniques. Firstly 50 photographs were taken with NIR camera and 830nm long pass filter. Afterwards, edge detection algorithms and morphological operations were performed on the images to obtain the boundaries of the images. Diameter and area information obtained from the binary image were used as attributes. These attributes were given as input to Linear Regression method and estimated. As a result, 93% of the diameters of the apples and 96.5% of the weights could be estimated.