Study of feature-based image capturing and recognition algorithm

In the field of applied computer vision, the three-dimensional (3D) object recognition is a very important technique which can be used to determine objects from a certain direction of the image for arbitrary 3D objects. These skills are useful in military applications, such as moving target recognition and coastal surveillance. Computer vision recognition allows fast response and all day long reconnaissance. The purpose of establishing a ship recognition system is to research and develop effective ship contour capture in the natural sea environment. We can develop a ship recognition system that is reliable and fast based on a database of ship images. We propose a recognition algorithm for ship images. This system utilizes gradient vector flow to capture the ship image contour first and calculates the geometric eigenvalue using this contour and Fourier descriptor. The eigenvalues are used to perform separate rough and detailed recognition. A graphic user interface is developed and the validity of the proposed technique is demonstrated using identifying images.