Research On Improved Algorithm Of Image Defogging And Visibility Detection Based On Wavelength Compensation

In foggy weather,a large number of suspended particles in the air scatter and attenuate the light on the surface of the scene.The higher the fog concentration is,the deeper the depth of field of the image is,and the more obvious the degree of attenuation is.The visual system has a poor effect on image feature extraction,which brings security risks to human daily work and life.Therefore,it is of great significance to remove the fog from the foggy image and restore the clear Fogless image so that the image can meet the needs of users.At the same time,it can quickly judge the visibility from the foggy image and timely monitor the early warning,which can effectively reduce the probability of safety accidents.In recent years,many scholars have done a lot of effective work based on the dark primary color demisting algorithm,and achieved good demisting effect,but after restoration,the color of the image will produce distortion,especially for the sky or white objects and other high brightness areas,and because the influence of visibility is not fully considered,the unified demisting method is adopted under different visibility Big error.At present,visual inspection mainly includes visual inspection and instrument measurement.The visual inspection is restricted by human psychology,perception and other factors.Instrument measurement is expensive and can not meet the needs of large-scale application.In view of the above shortcomings,the main contents of this paper are as follows.(1)When the dark primary color algorithm is used to defog a single foggy image,because the same transmittance is used in different areas of the whole image,especially in the restoration of the bright area of the sky,there will be more serious color offset problem.A wavelength compensation defogging algorithm based on different light is proposed.In this algorithm,firstly,the corresponding dark primary color image of the image is processed by median filter to eliminate the halo effect;then,the normalized values of the light attenuation coefficients of different wavelengths are calculated,and the different transmittance of theRGB three channels is obtained by using the different attenuation characteristics of the RGB three channels,and the atmospheric light value is modified;finally,the modified parameter values are combined with the atmospheric scattering model to get the image after defogging.Experimental results show that the algorithm can not only effectively remove fog and haze,but also reduce color offset,making the image more natural after defogging,especially for the fog image with bright areas such as sky.In view of the high cost of traditional visibility detection methods and the need for manual calibration,a new visibility detection algorithm based on the residual energy ratio of different wavelengths in the image is proposed.First,according to the guided filtering algorithm,the transmissivity map is obtained,and the visibility range of the image is preliminarily determined by the distribution of the transmissivity map;then,the normalized difference between the residual energy ratios of different wavelengths of the RGB three channels of the image is calculated,and the pixels with large gray value deviation of the single channel are filtered to improve the accuracy and operation efficiency;finally,the image visibility detection value is calculated.Experimental results show that the proposed algorithm can effectively reflect the visibility of a single image in fog,and it is simple and easy.(2)In view of the high cost of traditional visibility detection methods and the need for manual calibration,a new visibility detection algorithm based on the residual energy ratio of different wavelengths in the image is proposed.First,according to the guided filtering algorithm,the transmissivity map is obtained,and the visibility range of the image is preliminarily determined by the distribution of the transmissivity map;then,the normalized difference between the residual energy ratios of different wavelengths of the RGB three channels of the image is calculated,and the pixels with large gray value deviation of the single channel are filtered to improve the accuracy and operation efficiency;finally,the image visibility detection value is calculated.Experimental results show that the proposed algorithm can effectively reflect the visibility of a single image in fog,and it is simple and easy.(3)In view of the existing image quality evaluation methods rarely evaluate the degree of color restoration of defog image,and in the field of image defog,the problem of color offset of image restoration results is very common.A new image quality evaluation index based on different color light wavelength attenuation and visibility is proposed.The experimental results show that this method can not only reflect the effectiveness of different image defogging algorithms,but also evaluate the degree of color difference restoration.