Image Stitching Technology Research Of Gun Barrel Bore SPECIAILIZATION:Control Science And Engineering

The flaw detection of gun barrel bore is an important part of the maintenance of artillery. Panoramic images obtained through the gun barrel bore detection system, which can provide an effective reference for artillery weapon system design and maintenance, have an important guiding significance. The gun barrel bore detection system usually combines photoelectric detection and electromechanical drive, and gun barrel bore panoramic image is obtained though the late data processing. Currently, segmented image acquisition and stitching of the gun barrel bore is a hot research in the gun barrel detection field.In this paper, an improved image stitching algorithm is adapted to develop stitching study based on the acquired image data and complete the gun barrel bore panoramic images. Firstly, mathematical model of the optical detection system and image stitching model is built by analysis of self-designed gun barrel bore detection system. Secondly, considering more points mismatch defects for existing features by matching algorithm, clustering algorithm based on single-class and multi-class clustering feature points matching method is proposed to improve the image registration accuracy and stability. Then a nonlinear model combined with projection matrix and simplifies cubic curve model is established. A non-linear image stitching algorithm is proposed based on the different position, eliminating the image distortion interference caused by vibration during the detection system moving process. Effective implementation of the deformed image stitching is finished. Finally, by experimental studies, the images of a certain type of gun barrel bore are stitched and the panoramic images without significant distortion is completed.Through this study, the registration algorithms based on feature points clustering and location-based nonlinear image stitching algorithm are used to get the gun barrel bore panoramic images and provide data support for flaws detection of gun barrel bore, which have made effective results.