Abstract: Aiming at the problem of low accuracy in baggage shape and position measurement in the automated process of airline checked baggage palletizing, a method for measuring the size and position of airline checked baggage based on RGB-D images is proposed. Firstly, the image and point cloud data of the baggage are resolved from the RGB-D image based on the internal parameters of camera, the 3D point cloud and 2D image data of the main part of the baggage subject are extracted through clustering and perspective transformation. Then, the baggage rotation angle is measured based on the extracted baggage image and point cloud data. Finally, the optimal rotation angle is selected for baggage size measurement based on the surrounding quality of the bounding box. The experimental results show that comprehensive utilization of baggage image and point cloud information in RGB-D can effectively improve the accuracy of measuring the shape and position of baggage. Compared to that of the measurement simply using point cloud, the average error of baggage size measurement is decreased by 21.11%, the average error of position measurement is decreased by 11.80% and the average error of rotation angle measurement is de creased by 6.09%. It has achieved high-precision measurement of airline checked baggage and met the require ments of automated baggage palletizing process for baggage measurement

Key words: baggage measurement, machine vision, RGB-D image, edge detection, point cloud clustering