Navaphattra Nunak and Taweepol Suesut

pp. 629 - 633

Abstract

This paper proposes a new technique to measure the geometric mean diameter (GMD) of selected fruits and vegetables calculated from a three-dimensional (3D) image by computer vision system (CVS). From a single view of the image data a linear laser light projects onto the top of the sample through the center in order to mark the measurement points. The planar metrology and the measurement between planes are employed to calculate the width and height of the samples. Homography transformation and cross ratio are the mathematical parameter applied to calibrate the image data to real world distance (in metric system). GMD of sample can be calculated from a single view of the image with this technique. The percentage of error of GMD obtained from CVS compared with GMD measurements using vernier calipers is approximately 0.03-5.14
depending on the shape of the objects. However, it can be concluded that this technique is worthwhile for measuring GMD of symmetrical objects.

Keywords

• computer vision
• geometric mean diameter
• fruit
• vegetable