Image Acquisition Rate Control Based on Object State Information in Physical and Image Coordinates

Recently, three-dimensional (3-D) vision applications play an important role in industry and entertainment. Among these applications, interactivity and scene reality are the two key features affecting the 3-D vision performance. Depending on the nature of the applications, the interactivity with the users (or the players) can be regarded as an important need, for example, in the applications of 3-D games and 3-D virtual reality. On the other hand, 3-D vision scenes can be generated based on physical or virtual models. In order to improve the scene reality, a new trend in 3-D vision system is to generate the scene based on some physical or real-world models. For the 3-D reconstruction, research topics mainly focus on the static cases and/ or the dynamic cases. For the static cases, the reconstructed scenes do not change with the observed scenes. For the dynamic cases, the reconstructed scenes change with the observed scenes. For both static and dynamic cases, the 3-D positioning of moving objects in the scene is the major component of obtaining the object states in the physical coordinate. A typical 3D positioning system is shown in Figure 1. Cameras shown in Figure 1 are used to capture the images of moving objects in the physical coordinate. The image acquisition rate of Camera X is denoted as Rate X. After one camera captures the images using its designated acquisition rate, the captured images are stored and analyzed in the computer. After analyzing the images, the object states, e.g., position and velocity, in the physical coordinate can be estimated by any 3-D positioning method. The 3-D positioning plays an important role of obtaining the object states in the physical coordinate. In general, cameras are designated as the input devices for capturing images. Fundamental components of 3-D positioning are shown in Figure 2. First of all, images are captured by the input devices such as cameras. Before the image acquisition, the camera calibration should be done. Based on the calibration result, the relationship between the physical coordinate and the image coordinate in the captured images can be obtained. Since the quality of the captured images may be influenced by noise, some image processes should be performed to eliminate the noise. After the image processing, the captured images can be analyzed to extract the object information for further 3-D positioning. Finally, any 3D positioning method is used to find the object states such as the position and/ or velocity in the physical coordinate. O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m