Aerial Image Correction And Edge Detection:Application To Image Mosaic

The technology of image mosaic is to synthetize multiple images into a large scale and high resolution image,which is used in many fields,such as post disaster reconstruction,real-time map construction in wartime,agricultural monitoring and so on.However,due to the influence of flight environment,there are geometric distortions in aerial images.The existing step by step correction method increases the reconstruction error.Moreover,the correcting of image will produce noises in the edge region of the image,and make the shape of the effective scene of the image no longer regular,which leads to the obvious stitching marks in the image fusion.For these reasons,this thesis first improves the aerial image correction method.Then a solution is proposed to remove the edge noises of the corrected image.Finally,it is combined with image fusion techniques for image mosaic.The main work is as follows:(1)A direct conversion method for aerial image geometric correction without ground control points is proposed.Firstly,relative coordinate systems and the conversion relation between them are introduced.Then,the cause of geometric distortion of aerial image is analyzed using the collinear equation.Then,according to the flight attitude data such as pitch angle,roll angle,yaw angle and altitude,the camera imaging geometry model is calibrated.Finally,through the contrast experiment,three convolution gray interpolation method is used to resample the original image and construct the corrected image.The main contribution is that the geometric correction model is reanalyzed,the direct conversion relation of the corrected image to the original image is calculated,and the error caused by the reconstruction of the intermediate step correction interpolation is omitted,and the speed of geometric correction is increased.Experimental results show that the matching effect is improved while the matching speed is guaranteed.(2)A method of image edge detection for corrected images is proposed.Firstly,three forms of region selection are analyzed and compared.The scheme of selecting the maximum effective region and using the edge detection method to remove the noisy edge is determined.Image edge detection method involves two stages and uses a window function to detect the effective scene area of the image.For the first stage,the window function and threshold technique are used to distinguish effective scenes(with noise)and invalid scenes.The second stage uses the same window function and combines the predefined criterion to detect the noisy area with the edge of the effective scene(with noise),to locate and remove the noise to get the noise-free effective scene area.Finally,the experimental comparisons with a simple overlaping fusion are carried out.The results show that there is no obvious stitching trace in the mosaiced image by using the image edge detection,and the quality is obviously higher than that omitting image edge detection.This shows the effectiveness of the image edge detection method.(3)The image edge detection method is combined with several image fusion techniques to stitch aerial corrected images.Firstly,the image edge detection is performed on the overlapping area of the corrected images,and a two-value map corresponding to the effective area is obtained.Then,according to the corresponding relationship between the two-value maps in space,the overlapping area of real effective scene is determined.Finally,several image fusions processes are implemented on the overlapping area to achieve image mosaic.The experimental results show that this method can not only solve the problem caused by the noisy edge of the corrected image,but also solve effects of the invalid pixels filled in the corrected image to image fusion and stitching,and reduce the influence of the stitching trace in overlapping area,and smooth the transition.