The Research On Dodging Technique Of UVA Image

The uneven illumination of UAV image is the phenomenon of uneven distribution of brightness and contrast in the image.The dodging technology is the theory and method of solving the problem of uneven illumination.The three forms of the uneven illumination phenomenon of remote sensing image:uneven brightness,uneven contrast or brightness distribution uniformity and contrast of the uneven distribution exist at the same time,the dodging algorithm of UAV image is studied in this thesis,the main work is as follow:1.The present situation of the research of the dodging algorithm at the domestic and foreign is analyzed.And after experiment,The characteristics and applicability of the existent dodging algorithm is analyzed and summarized.2.For solving the problems of how to choice the best size of Gauss filter of MASK dodging algorithm,and the contrast is inconsistent in the image after MASK dodging processing.A block MASK dodging algorithm based on distance weighting was proposed in this thesis.Due to the various regions of an image have the different definition,the size of the Gauss filter is not the same,the background image that fit the whole image is difficult to be obtained by a single size of the Gauss filter.The improved algorithm is proposed in this thesis,based on distance weighting block mode,the cut-off frequency of each block's Gauss filter is determined by the image signal-noise ratio,the better quality background image is obtained.Finally through the gray value of the background image on the subtraction image make the different degree of tensile contrast,gaining consistent and brightness balanced image.3.A dodging method of UVA is proposed in this thesis by the characteristics of large scale,high precision and high quality of the UAV image.Because the essence of MASK dodging algorithm is to make the brightness of the region to weaken,to enhance the brightness of the low brightness area in order to achieve the effect of brightness balance.However,the highlight area of the UAV image is usually normal brightness,and don't need to be weakened.After the area of UAV images after MASK dodging processing prone to ground color distortion and other issues.To solve these problems,a dodging method is proposed in this thesis that brightness compensation low light area,through the Gauss filter to extract the low light area and brightness compensation coefficient,and the compensation parameters are obtained by the brightness mean value of the highlight area,to make the whole image to achieve brightness balance effect.The experimental result that the method is more suitable for UAV image,and can avoid the ground color distortion of the highlight area.Finally,the optimal process of UAV image dodging is determined through experiments.