| Airborne optical measurement system is a new application form of optical measurement, which is flexible, real-time, high-efficient, and possesses wide measuring range. Aiming to develop video processing software for airborne optical measurement system, feature based object detection and tracking methods have been studied in this paper, and video processing software for airborne optical measurement has been designed. The main work of this paper includes:1. Response characteristics under airborne optical measurement situation of different feature detection algorithms, such as SURF, SIFT, MSER and FAST, have been analyzed and compared, and SIFT has been chosen to be the main algorithm in this paper.2. Difference image method for moving object detection based on feature matching has been introduced. In order to eliminate the influence brought by camera motion more efficiently for situation under which the relative movement of object is small, feature selection method using local gray scale of difference image has been presented.3. Object detection and tracking method based on implicit shape model has been researched, original method has been proven to be unable to estimate the theoretical defect of each interpreting point, and the error expression has been deduced. According to error mechanism, dual training method and method based on main direction of feature have been proposed, which improve the detection and tracking precision.4. The system design, hardware construction, software composition and key technologies of airborne optical measurement video processing software has been introduced in detail.The object detection and tracking technology proposed in this paper can be further applied to many other domains, such as airplane passive location, precision guidance, battlefield surveillance, security control and intelligent transportation, etc. |
PDF Full Text Request