Surface area and volume measurement using radial projections

Abstract

Quality evaluations in agriculture and food processing, and medical imaging often require accurate non-destructive surface area and volume measurements. A computer vision technique has been developed to achieve this task for objects with irregular shape. An imaging system was built to acquire and store images of multiple projections of the objects. Each of these images was taken at the same angular interval. Image segmentation and tracing algorithms were implemented to provide the x and y coordinates of the boundary points for each image projection. These boundary points were then used to generate a three-dimensional wire-frame model. The result of surface fitting and approximation on the wire-frame model was used to calculate the object surface area in this research. Object volume, although not covered in this research, can also be measured. Examples of agriculture and food processing applications using this vision system for surface area measurement are included in this paper. Two major challenges of this research work, system calibration and surface approximation, are discussed in this paper.