Study On Fog Removal Algorithm For Single Forest Images

Due to the existence of natural fog,the image acquisition of the forest monitoring system based on computer vision has a serious impact,so that the acquired forest image is degraded and degraded.In order to minimize the influence of fog or haze on the image,the fog is related to The algorithms are endless.The proposed method is more targeted and specific than other defogging algorithms.It thoroughly studies forest images in fog,and makes the overall image of fog forests clear and efficient.It is applied to real-time monitoring equipment for forest monitoring and detailed identification.This has high requirements on the computation time and processing accuracy of the algorithm.This paper studies the effectiveness of the algorithm based on both model and non-model directions.The main research work is as follows:Firstly,the fog-day degradation mechanism and the current image enhancement methods are analyzed.Based on this,the improved histogram equalization algorithm and the Retinex method are proposed.The algorithm idea is:firstly,the global contrast in the fog is increased and enhanced by the CLAHE algorithm.The details of the image are then segmented and extracted from the sky area where the processing is not effective.The sky area is processed separately and the image enhancement is performed using the improved Retinex algorithm.Experiments show that this method solves the traditional image enhancement algorithm in dealing with forests in fog.A series of problems such as color distortion and halation in the image have a very good effect.Then,based on the effectiveness of the model-based dehazing algorithm applied to the forest image in fog,several popular methods of defogging were analyzed.Based on the priority of dark channels proposed by He,bilateral filtering is used for edge preservation and defogging,and bilateral filtering is used.The method indirectly solves the transmission of the medium and improves the transmittance and the atmospheric light.Through experiments,the improved method not only achieves a good defogging effect,but also keeps the details of the image clear.At the same time,the operation time is shortened and reduced.The complexity of the algorithm is more suitable for real-time monitoring of forest images.This article uses a combination of subjective evaluation and objective evaluation(mean gradient,time)to evaluate the defogging effect of forest images.The effect maps processed by the two improved methods are compared with traditional methods.The results show that The method proposed in this paper has a good effect both objectively and objectively.After experimental verification and subjective and objective analysis,both types of algorithms can achieve certain effects in recovering the fog-free forest image.This paper has solved the dehazing method with histogram equalization and Retinex algorithm from the perspective of non-modeling algorithm.The sky block effect and color distortion problem;and in the model-based algorithm perspective,the improved dark channel priority algorithm solves the He algorithm's ineffectiveness in the sky region,handles the nature,and restores the details of the image well.And greatly increased the operating speed.