Research On Single Image Dehazing Algorithm

Fog is a common natural phenomenon,which is prone to appear in some southern places of China,the northern regions are also easy to produce in autumn and winter.In the fog,the air contains a lot of aerosols,which lead to the scattering on the scene light,causing the contrast,saturation and color characteristics of the captured image have attenuation in some degree.Narrowing the dynamic range of picture color and losing details,all of which have a direct impact on the work of outdoor visual system,such as outdoors monitoring system,the snapping system and the aerial photographing work.Affect people's further research and application to the acquired images seriously.There are some defects of the foggy image,which contain the contrast decreased and the details lost.With the degree of the field depth increasing,the fog becomes thicker.By studying the imaging mechanism of foggy images,we will present a new dehazing method in this paper.Different from the previous methods based physical model restoration or based image enhancement,our algorithm based on the idea of the image fusion,using the new fusion method to recover the images obtained in the foggy day.Specific research methods are as follows:First of all,from an intuitive point of view,to dehaze is to remove the veil between the scene and the imaging sensor.It is true that the veil will grow thicker with the depth of the increasing field.Therefore,in order to make the defogging effect more realistic and natural,it is necessary to obtain the depth map of the image.Used the edge-preserving feature of the tree filter,the dark channel map can be optimized to an approximate depth map.Combining the depth map,we remove the fog veil,making the image's color saturation and contrast enhanced.Second,as the overall brightness of the haze image is bright and the image contrast is low,Gamma correction is performed on the original foggy gray image.By Gamma correction,the overall brightness of the image is reduced,while the contrast at the high brightness regions of the image is increased.The details information of the fog image are lost,we use local contrast maps to enhance the details of the Gamma corrected image.Finally,due to the problem of over-saturation of the image obtained in the first step,and the image acquired in the second step is a gray image,it is necessary to apply the image fusion.After fusion,the image is defogged well.The experimental results show that the restored images obtained by the new algorithm have high contrast,clear texture and abundant details,and there is no color distortion in the sky regions.