Volume measurements using RGB cameras

Abstract

The following thesis explores the possibility of an inexpensive system using two RGB cameras to measure the volume of objects. Measuring volume is a critical task in the logistic industry as the price of moving objects is often connected to the volume. Many smaller locations lack automated systems and rely on time-consuming manual measurement. The proposed system provides an inexpensive way to measure the volume automatically. It also provides tracking possibilities by saving images of measured objects.To achieve the objective the vision area of calibrated cameras was explored and calculations used to show that it is feasible to use computer vision techniques to measure volume from pixel data to real-world dimension. A prototype system was built and tested on packages ranging in size from 5 cm to 30 cm. Further, an algorithm utilizing computer vision and the OpenCV library was developed to extract contours and determine the pixel size of objects. With the pixel size of the objects and calibrated cameras the real size of the objects was calculated. Results demonstrate the feasibility of using RGB cameras for accurate volume measurements, achieving an average error of 0.66 cm. The system performs well under controlled lighting conditions, provided the object’s color contrasts with the measuring area. While promising further development is needed to present a fully functional device. Automating the calibration process and enabling real-time operation are key next steps. The following study highlights the potential for low-cost, camera-based solutions to enhance operational efficiency in small-scale logistics applications.