Key Technology Of Attitude Estimation Using Afterglow Effect Processing Of Downward-looking Scene Image

Autonomous flight is one of key technique in UAV practicality, for the affect of environment andother factors, GPS signal is sheltered and lost easily which makes attitude measurement of integratednavigation system used by UAV drift and divergent gradually. Establishing and utilizing UAV visualsystem provides navigation control for steady flight.The paper estimates the attitude of aircraft based on afterglow effect simulation, the main researchof the paper is as follows considering the key technology among the field:1) elaborate imaging principle of the sensor in UAV remote sensing system and mapping betweenflight attitude and external azimuth of senor. On the basis of establishing the imaging mathematicalmodels of downward-looking scene image, analyze the influence of attitude on imaging region.2) a method to simulate afterglow effect on attitude aberration of downward-looking scene image isdissertated. Simulate aircraft in uniform speed with different attitude to get image sequence withgeometric distortion, and image afterglow effect by smoothing filtering in time domain to appear thatgray level of downward-looking scene image pixels when senor moving will keep for some timebefore disappearing. Extract afterglow lines which are in the edge of several regions whose grey levelvaries obviously, to get the relationship between afterglow distribution and UAV attitude according tofixed law.3) proposes a method of attitude estimation based on afterglow effect simulation. According to thepattern of afterglow lines with attitude change, select proper characteristics of afterglow lines,discussing the way to estimate UAV attitude individually. Estimate yaw angle with the mapping to tiltangle. For pitch angle and under the circumstances of pitch and yaw angle, introduce image jacobianand simplify senor system to uncalibrated visual servoing system to excursive angle cyclically. Whenroll angle is also not zero, combine image jacobian with feature points to estimate attitude.4) design GUI, import video to make an estimation of attitude.Firstly, preprocess the aerial video.Then integrate the algorithm and simulation to design interface and functional modules. According toconstract of the base image (attitude angles are zero) and geometric correction of real-time image toverify the proposed method, the result of simulation proves reliable.