Research On Target Detection And Location & Orientation Estimation

Aerial refueling technology, which is hailed as the multiplier of air power, is widely used in modern war and studied all over the world. In the complex atmospheric environment, it highly risks to dock the probe and drogue precisely. Thus, developing a guidance system of assistant aerial refueling is urgent and necessary. This thesis will research on the contour extraction and estimation of the three dimensional location and orientation of the target taper sleeve based on real-time image sequence.Firstly, the thesis will research on the technology of image preprocessing. It will introduce two color modes of RGB and HSV, then it will give their conversion relationship. After that it will discuss the image enhancement method of Retinex and give two algorithms of center surrounding and globle Retinex. Because of two problems on pixel gray value correction and stretching transformation, it will propose a revised globle Taylor Retinex algorithm. Later periods processing of Retinex enhancement, Otsu threshold segmentation and morphological closing filter will be used to enhance the image.Secondly, it will research on image contour detection. It will introduce Hough transformation and RHT method by discussing two methods of edge tangent and symmetry, and then because of problems on huge computation and failure situation, it will propose a revised detection algorithm based on characteristic edge points. Later periods processing of contour extraction, long straight lines removal, edge points linking, line segment fitting and ellipse contour detection will be used.Furthermore, it will research on 3D estimation of location and orientation. It will introduce the monocular visual measurement including the coordinate system establishment, rigid transformation, affine and perspective transformation. Then it will discuss methods of single or concentric-coplanar circular features, and based on the characteristic of the taper sleeve mode, it will propose a revised estimation algorithm based on feasured diameter, which comfirmed validity later.Finally, the system of real-time image collecting and processing are built. Under the influence of changing external environment, light condition and motion status, it will test the performance of the system. The test result shows that the system is effective, stable and precise to detect and calculate the location and orientation of the target taper sleeve.