Research On The Image Dehazing And Matting Algorithm

The images matting refers to the problem of decomposing an image into two layers,called foreground and background,which is widely used in image editing and film production.However,the haze in China is frequent,and the haze adversely degrades quality of an image thereby affecting its aesthetic appeal and visibility in outdoor scenes.Such images are difficult to apply directly in both foreground deduction tasks and other visual processing tasks.Most existing work on dehazing,including the recent convolutional neural network(CNN)based methods,rely on the classical mathematical formulation where the hazy image is modeled as the superposition of attenuated scene radiance and the atmospheric light.In this work,inspired by the Densely Connected Pyramid Dehazing Network(DCPDN),we propose a new multi-scale image dehazing method using Perceptual Pyramid Deep Network based on the recently popular residual dense blocks to directly learn a non-linear function between hazy images and the corresponding clear images.The proposed method involves an encoder-decoder structure with a multi-scale pooling module in the decoder to incorporate contextual information of the scene while decoding.Extensive experiments demonstrate that the proposed method achieves significant improvements over the state-of-the-art methods including the Densely Connected Pyramid Dehazing Network(DCPDN).Our method has a smaller network model than the Densely Connected Pyramid Dehazing Network.In the same PyTorch operating environment,the network model size is 9.8MB,which is only 3.83% of the DCPDN.And experiments on 800 indoor and outdoor images shows that our algorithm has a better performance on both peak signal-to-noise ratio(PSNR)and structural similarity(SSIM).Inspired by the Bayesian matting,we propose a quick bilateral filter matting method in Bayesian framework.The algorithm adds a color correlation constraint to optimize the sample point weight.this paper we combine the bilateral filter with the Bayesian matting,and uses the bilateral filter to optimize the complicated process of the original algorithm.Finally,the experimental results show that our algorithm based on bilateral filter under the Bayesian framework proposed in this paper has less statistical error than the original algorithm on 27 authoritative data sets,and reduces the average of each picture by 56.24% of Processing time.In recent years,with the rapid development of social networks and smart devices,the image matting is used wildly.On the other hand,the haze adversely degrades quality of an image thereby affecting the visibility.The methods proposed in this paper can solve the problem of image degradation in outdoor scenes and the image matting.