In tracking maneuvering targets, ascertaining the start and the end of a maneuver time is a prerequisite to maintaining good track quality. Existing maneuver detection schemes based on point-mass tracking are either inadequate to reliably detect rapid maneuvers or computationally impractical. Maneuver detection based on images can be more accurate but is also very demanding on computational and storage resources. A new method of detecting maneuvers using information from an image sensor is presented. It involves the use of two detectors, one to detect linear acceleration and the other to detect orientation changes of the target. To detect target maneuvers, the detectors continuously monitor the position of three points extracted from the target image-the centroid, the front centroid, and the rear centroid. Contrary to existing image-based tracking methods, the approach put forth is very attractive from computational and storage points of view. The proposed method was tested using simulated tracks and the results confirm the benefits of its utilization for maneuver detection.