Search

In this talk, we give an introductory overview of characteristic modeling and characteristic model based control. First, we review its basic concept, key features, and underlying mechanism. Then, we overview the controller design process based on characteristic model. In order to demonstrate the practicability of characteristic model and the superior performance of the closed-loop control system under characteristic model based control, we introduce its successful applications in several engineering problems, and in particular we provide in detail its applications in the automatic rendezvous and docking between Shenzhou spacecraft (including Shenzhou-8, Shenzhou-9, and Shenzhou-10) and Tiangong-1. Thirdly, we review the existing theoretical results on characteristic modeling and characteristic model based control. At the end of this talk, we discuss several future research topics and prospects concerning characteristic modeling theory.

Top-hat transformation has been widely used for infrared small target detection, but it is sensitive to the clutter and may destroy the details of the image, which make it an ineffective way for small target detection. To detect the infrared small target simply and efficiently, a new Top-hat transformation based on contour structuring element is proposed and then modified according to the property of the target region to enhance and detect the infrared small target in this paper. Experimental results indicated that the proposed method was efficient under the conditions of heavy clutter and dim target intensity.

Multi-target tracking is a crucial but complex problem in small target tracking applications, especially when large number of false alarms existed. In order to simplify the multi-target tracking, a novel simple algorithm is presented by using motion properties to reduce false alarms, which is implemented through mathematical morphology and multi-frame cumulation. Firstly, the consecutive images are dilated and cumulated according to the motion properties of the small targets, thus generate the cumulating image. Secondly, most of the false alarms are reduced from the cumulating image through mathematical morphological operation. Finally, the resulting consecutive images including targets and very few false alarms are restored from the processed cumulating image and the original consecutive binary images. Experiment results illustrated that most of the false alarms were excluded and the small targets were easier to be identified, which largely decreased the computational time of target tracking algorithm, and thereby decreasing the complexity of the small targets tracking.

The imaging model of an optical camera without image aberration can be equivalently depicted by its conjugate nodal plane model. A pinhole camera model itself neglects aberration of corresponding equivalent model, and also fails to represent shift error of flat glass, tilt error of optical axis as well as image aberration of an actual optical system, all of these four factors give rise to image distortion. Based on this, utilizing the property of conjugate nodal plane, an geometry model of image distortion is presented. This model, which fits for pinhole cameras with chromatic aberration corrected, can analyze symmetric and asymmetric image distortion. The simulation of an actual optical system indicates that the relative radical distortion and relevant angle error both calculated by this model appears in the range of rational design parameters. In this model, four unknown parameters needed to be recognized are principal relevant parameters, tilt angle of optical axis and axial spherical aberration of flat glass. Theoretically, this geometric model can be used to correct image distortion, aiming at better precision of calibration of image distortion.