Offline Stereo Camera Calibration of Raspberry Pi Compute Module. Abstract- Computer vision has become a very popular field due to its numerous applications. Stereo imaging, one of the areas of computer vision is frequently used in many applications like autonomous robots to calculate the free path, 3D reconstruction of environment, automatic cars, travel aid for visually impaired and many more. The major step in stereo imaging is establishing correspondence between multiple images. Errors within the camera and the method of arrangement of cameras can lead to wrong correspondence. Hence the depth information calculated will be incorrect. To increase the accuracy of depth matching in stereo imaging, we need to correct the camera imperfections. Stereo camera calibration is a method that can help predicting the internal and external parameters of the camera. There are many methods to perform camera calibration. In this paper, we explain the method of calibrating stereo raspberry pi cameras connected to raspberry pi compute module using 2D calibration object.
Keywords: Computer vision, stereo imaging, depth mapping, stereo calibration, raspberry pi, compute module.
INTRODUCTION
Computer vision is an important field of research. Computer vision technique helps to model complex visual environment using various mathematical models. A stereo vision is an important field within computer vision that addresses an major research problem of reconstruction of three dimensional coordinated points for depth estimation [1]. A system of stereo vision consists of a stereo cameras placed horizontally. The two cameras capture images simultaneously and the captured images are processed for recovery of visual depth information. Depth recovery depends on the concept of establishing correspondence between pixels of images captured by stereo cameras. Cameras internal parameters and alignment of cameras with respect to world coordinates govern at which pixel the image of a point on the object in focus will be mapped.