主動式視覺系統之動態追蹤

Abstract

影像系統常應用於自動駕駛、監控系統、以及保全監視系統上。在傳統的應用中，影像系統扮演被動觀察者，然而，人類在觀察物體時，總是能夠主動調整好視角，將鎖定的物體置於視野中心使之容易辨識，同時物體也不易離開視野，所以主動式的觀察者較符合人類的視覺感覺，同時可以預先補償延遲的現象。主動式視覺系統乃是架構在射控系統上而其結合兩種子系統:影像處理系統負責動態偵測，找出變動的物體，另一個資料庫系統將根據之前的動態偵測更新物體的狀態，來達到主動式的視覺，而主動式的視覺系統與應用在軍事上的射控系統相結合將可以達到動態追蹤。

Image-based systems are utilized extensively in the automated guidance, surveillance, and monitor. In the traditional applications, image systems play the passive roles recording what they catch. However, the visual perception of human is active. The sights of humans are always adjusted to make better views of the targets. Humans also memorize the targets so that they follow them constantly. The hardware of active vision system relies on the sighting fire control system. The thesis illustrates the method to implement the active vision system based on the human perceptions. Active vision systems consist of two parts: the motion detection module and the object database module. The former system detects the motion differences and finds out the schemas of the objects. The later system records the objects and updates the object data according to the motion detection of image processing system. By combining the two portions, an active vision system is achieved. The sighting fire control system will play the role to track the target that exists in the object database. Therefore, the integration of active vision system and sighting fire control system can realize motion pursuit.