![]() ![]() Because the algo- rithm is nonlinear, Bode plot cannot be used. A bandwidth evaluation can be conducted by applying the VSDC algorithm to track sinusoidal signals. In order to fully utilize the capability of the camera while avoid queuing the image processing tasks, the sampling interval of the camera is set close to the time required by image acquisition and processing of a single frame. This is because for industrial machine vision systems, usually the minimum sampling interval of the camera is shorter than the time consumed by image acquisition and processing. Maximum latency is assumed, i.e., the image acquisition and processing actions are assumed to occupy a whole time step of camera sampling. Figure 4 demonstrates the effect of the VSDC algorithm. Because of the sensing latency, the correction action is applied to a past state, whereas the present state is predicted from the corrected past state. ![]() As ex- plained in Fig.3, the correction action is conducted every N prediction steps. ![]() A dual-rate Kalman filter can be applied to Eq.(4). The value of its covariance W is adaptively determined using a learn- ing algorithm. w is the equivalent noise that replaces the actual input u in Eq.(2). A, B w, and C are as shown in Eq.( 2) and (3). ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |