关键词：传感器；成像；车载无人飞行器；数学建模
摘 要：The implementation of imaging systems onboard UAVs allows the development of original methodologies for extracting information concerning the surrounding environment, as well as of the vehicle itself. Imaging sensors, are very attractive since they are passive, non-contact, very versatile, and low-cost. In addition, they can be used in situations where other sensing devices fail, leading to a whole new group of applications that can be accomplished. However, before using imaging sensors a mathematical model must be computed, describing how the 3-dimensional points are represented in 2-dimensional images. For the interested reader, there are numerous computer vision books covering related topics in great detail, see for example [40] and [60]. In the present chapter the required theoretical background for the construction of imaging models is briefly presented, and in addition, the physical implementation of the imaging system for estimating the sates of a UAV is also addressed. This chapter is divided as follows. Section 5.1 presents the pinhole camera model, as well as a camera calibration procedure for obtaining the intrinsic parameters. Next, stereo imaging is introduced in Sect. 5.2, with an explanation of the epipolar geometry concept. Also, a method for stereo calibration and rectification is presented. With the purpose of allowing the estimation of relative translational speed using an imaging sensor, the concept of optical flow and a method for its computation are detailed in Sect. 5.3. In Sect. 5.4, some important points that must be considered when implementing an imaging system onboard a quad-rotor UAV are discussed. In addition, the development of both a monocular and a stereo imaging system is presented, as well as the software architecture conceived with the purpose of estimating the data required for performing vision-based tasks. Finally, some concluding remarks are presented in Sect. 5.5.