ABSTRACT
The very first step of most real-world visual computing applications is the acquisition of images or video. However, acquiring meaningful image and video data is surprisingly challenging. This stems from the fact that realworld cameras and sensors are far from perfect concerning sampling or measuring and a substantial amount of processing needs to be applied before the data can be used. The different sensor types will be discussed, in particular with regard to how they affect the quality of data, how measurement noise affects image acquisition, and how to create dense color samples from sparse samples as they are acquired by most cameras. In order to characterize a given camera, it has to be calibrated in terms of radiometry and color as well as lens and geometric calibration. Applying all these steps results in well-calibrated and meaningful images.
