Research On Image Dehazing Algorithm Based On Laplacian Pyramid

In foggy weather,various suspended particles in the air will cause severe atmospheric scattering.At this time,the main color of the picture taken by the imaging device is gray and the quality is poor.This image is not only poor in visual sense,but also unfavorable to the computer.Identification.Therefore,there is a clear practical significance for the dehazing treatment of foggy images.At present,the mainstream processing methods are based on the dark channel prior principle.This type of method can complete the defogging of most images,but the image processing results for areas with large areas of the sky are not good.In this paper,the principle of atmospheric imaging in foggy weather is analyzed firstly.The direct attenuation model and the atmospheric attenuation model are introduced.The advantages of the dark channel prior principle dehazing algorithm are analyzed.The algorithm is simulated by Matlab and the image transmittance is completed.The quadratic optimization results show that the proposed algorithm is not ideal for the fog removal of the image sky region,and it is easy to introduce new noise and image texture,and cause overall color distortion.Since the dark channel prior principle algorithm works well for the rest of the sky except the sky,the strategy of splitting the foggy image into the sky part and the rest is proposed.This paper optimizes a K-means clustering method to achieve Quick and effective regional division.For the sky part of the image,the gamma correction and image pyramid are used to defogg,and finally the image is fused.For the rest of the image,the dark channel prior principle algorithm is used to defogg.The gamma correction method used in this paper can enhance the image sense,but this method is easy to lose the edge information in the subsequent image fusion process.For this defect,the edge information protection feature of the Gaussian kernel function and the multi-scale and multi-resolution of the tower shape decomposition are used.The advantages of the rate to make up.The normal exposure image is prone to image overexposure after gamma correction.Therefore,this paper proposes to fuse the underexposed image and estimate the contrast and saturation as the fog map.The mapping weight is combined with the Gaussian pyramid and the Laplacian pyramid to complete the image.Fusion defogging.The algorithm is simulated on the Matlab software platform and compared with the dark channel prior principle dehazing algorithm.The results show that the algorithm achieves better dehazing effect,noise is eliminated,the output image contrast is higher,and the algorithm is robust.Sex is also stronger.