Automatic Battle Damage Assessment Of Bridges Based On Image Understanding

Battle damage assessment is a new research field, and is still on its initial stage. Thanksto its perfect future in application and importance in military affairs, battle damageassessment has been abstracted a great attention of researchers. This dissertation focuses onthe following aspects of battle damage assessment system for bridge (BDASB) based onimage understanding: (1) Abstraction and matching of the key position of the bridge; (2)Expert system of BDASB; (3) Design and implementationof BDASB. Firstly, the research background of BDASB is analyzed. The basic problem of BDASBand its key technology is analyzed, and a brief introduction of the main research work in thisdissertation is given. Secondly, the abstraction and matching of the key position of BDASB is studied. Thekey position characteristic of the bridge is presented by its contour. Abstraction method of thekey position contour for bridge using Canny edge detector is put forward, which can get thecontour characteristic of bridge effectively through setting the threshold value dynamically.Then the bridge image attacked is matched through the template matching using contourcharacteristic. Experiments demonstrate that this method can abstract and match the keyposition of the bridge effectively. Thirdly, the knowledge-based expert system of BDASB is studied. Knowledge-basedexpert system of BDASB is set up based on instance, and implemented using the relevantdatabase. When the rule information of the images of bridges is given, it can be matched tothe rules set in the knowledge base, and then the correct evaluation can be found. Theknowledge base is maintained through reasoning machine and database. Experiments showthat the expert system is quite efficient, flexible and fault tolerant. Finally the design and implementation of BDASB system is analyzed. The systematicsoftware prototype of BDASB has been set up. This system prototype is developed usingOOP, and canabstract the problem of BDASB effectively. Knowledge-based expert system isrealized by the relevant database, which can guarantee the completeness and consistency ofthe knowledge base. The systematic prototype utilizes the software project method, whichmakes it has good expansibility. The prototype can be expanded through addition andmodification of typical object, imaging sensor, object recognition algorithms, evaluationmethod, criterionet al.