Defogging Of A Single Forest Image Based On Threshold Segmentation And Dark Channel Prior Method

Due to frequent forest fires and deforestation,large areas of forests and grassland green spaces that can absorb carbon dioxide on the earth are disappearing,which leads to the imbalance of carbon and oxygen cycle and the greenhouse effect.Therefore,it is urgent to protect forest resources,among which it is particularly important to monitor the forest condition in real time,prevent forest fires or quickly obtain fire information at the first time when forest fires occur,so as to control the spread and extinguish the fires in time.However,the natural smog and fire in the forest produce a lot of smog,and the tiny particles mixed with it will scatter the atmosphere,which will affect the quality of the images collected at the scene.Images affected by smoke will make rescuers unable to judge the situation of the fire scene well,which greatly affects the efficiency of disaster relief.Therefore,it is necessary to develop an effective and rapid method to remove smoke,and to remove smoke from the images collected at the scene,so that rescuers can judge the situation of the fire scene according to the clear images,and thus take more targeted and effective methods for disaster relief.The traditional dark channel prior algorithm can’t deal with bright areas(sky,dew,etc.)in the forest,and the defogging effect of images with uneven fog distribution is not ideal.In order to solve this series of problems,this paper proposes a single image defogging method which combines threshold segmentation algorithm with dark channel prior algorithm.By preprocessing the image,the value of atmospheric light is optimized,and the experimental results show that,compared with the traditional dark channel prior algorithm,this algorithm not only has good defogging effect on forest natural cloud images,but also has good robustness on forest fire smoke images.The following are the main points of work of this paper:(1)This paper introduces various methods of image defogging and the development status at home and abroad,and further studies the threshold segmentation algorithm,which divides the image into bright areas and non-bright areas with smoke and flame as the boundary,and eliminates most of the noise in the image.Expand Otsu algorithm,select different Otsu algorithms according to different environments,and help Otsu algorithm calculate threshold quickly by combining with bimodal method.This method has a good segmentation effect and can reduce the complexity of the algorithm.(2)Compare with image enhancement and other methods,analyze the nature of image restoration,understand the atmospheric scattering model in depth,and divide the atmospheric scattering model into two parts: the incident light attenuation model and the atmospheric light model.At the same time,the foggy imaging model is introduced,the causes of fogging and its influencing factors are analyzed,and the causes of image degradation are studied.(3)The calculation details of the dark channel prior algorithm are analyzed in detail.According to these details,the improvement of the algorithm is analyzed,and the influence of two important parameters on the experimental results is illustrated with examples.The quadtree method is used to calculate the atmospheric light value on the image after threshold segmentation,which effectively overcomes the problem of erroneous estimation of atmospheric light caused by the existence of many different light sources in the image.Then,the transmittance is refined by conduction filtering,and the algorithm efficiency and calculation accuracy are improved.(4)Finally,the construction of the experimental environment is described,and some key codes are displayed.Compared with other defogging algorithms,the feasibility of this algorithm is proved from subjective and objective aspects,and the images in complex scenes still have strong anti-interference ability,and the defogging effect and computational efficiency of the algorithm are better optimized.