Visual odometer technology is a navigation mode using only scene image information to calculate the motion parameters, which has good development prospects in positioning and orientation of the autonomous intelligent vehicles. In this paper, the main advantages and technical characteristics of vehicle visual odometer system is summarized and the general architecture and application systems are introduced. The development and current status of the crucial technologies of the odometer, such as camera calibration, feature extraction/matching algorithm, and motion parameter estimation, are mainly analyzed. The implementation schemes for high-precision vision odometer system and high-speed visual odometer system are proposed, and the main algorithms involved are described in detail, which may be taken as a reference for visual odometer research.