Research On Image Dehazing Methods Based On Boundary Constraint And Contextual Regularization

In the foggy weather conditions, the quality of images acquired from the outdoorsignificantly reduced, image looks dim in color and the contrast reduced sharply, someimportant details are lost in the mist, which brought great difficulty to object recognition.It is hard to perceive detailed information in images, so the practical of outdoor videosurveillance system is affected greatly. Therefore, image dehazing became a keytechnology to enhance outdoor image quality.In order to solve the problem of artifacts in the field of image dehazing, theboundary constraint and contextual regularization based image dehazing method is studiedin-depth in this paper, the core of this method is the estimation of scene transmission.However, in the process of estimating on the scene transmission, its constraints are notalways met, the miscalculation of scene transmission may be happened in some areas, sothis method is analyzed and improved in terms of the miscalculation of scene transmissionin this paper. On this basis, a three-stage estimation method for scene transmission aredesigned: the first stage is the coarse estimation of scene transmission, a patch-wisetransmission is being got through the boundary constraint; the second stage is the fineestimation of scene transmission, an element-wise transmission is being got through thecontextual regularization; the third stage is the correction of scene transmission, a morerealistic scene transmission is being got through the tolerance mechanisms. Finally, in theframework of atmospheric scattering model, a clear haze-free image is being got throughthe real scene transmission.In this paper, a number of experiments on the image dehazing are presented foranalysis and comparison. The experimental results show that this method has a good effecton edge and details preserving, and can handle the bright sky region very well and alsoproduces fewer halo artifacts.