Effects of Camera Lens Distortion on Particle Size and Shape Characterizations

Abstract

Abstract The image-based methods have been widely used for particle size and shape characterizations. However, images captured by various cameras may contain a level of lens distortion. This study shows that even a small distortion that cannot be identified by visual observation could significantly alter image-based particle size and shape distortions. This study develops a self-rectification technique to rectify the distorted images. The key is to use a rectangular template as a reference system. An edge detection technique is used to identify the edges of the rectangular template, which becomes curves in the distorted images. Image processing techniques based on least square and Levernberg-Marquardt approaches are used to rectify the curved edges back to straight lines and determine distortion coefficients. The distortion coefficients are used to quantify the magnitude of image distortion and to rectify the image. This research demonstrates that it is unreliable to use visual observation to determine whether the image contains distortion. The rigorous analysis must be performed to ensure the accuracy of image-based particle size and shape characterizations.