Research On The Enhancement Method Of Single Backlit Image

Being widely used in the field of digital image processing and computer vision,images serve as carriers of information.Backlit environment is relatively common while shooting,therefore backlit images influenced by shooting environment occupies a large proportion.The meaningful area(dark area)of backlit images usually present low visual quality,unobvious details,and serious color distortion,while the background area usually presents overexposure,loss of detail,and low contrast.The above-mentioned problems greatly reduce the application value of backlit images.At present,there are few researches on the enhancement processing of backlit images.Existing enhancement algorithms often lead to insufficient enhancement of dark areas,over-saturation of bright areas,and loss of details and color information.The enhancement effect is not very satisfactory.Based on the above factors,this paper continues to conduct in-depth research on backlit images enhancement processing.The main research contents are as follows:Aiming at the problems of insufficient dark area enhancement and oversaturation of bright area in existing algorithms,a backlight image enhancement algorithm based on virtual exposure method is proposed.Firstly,we introduce virtual exposure images to combine the characteristics of backlit images and the exposed images.Low exposure images are used to represent the bright area and high exposure images are used to represent the dark area,which improve segmentation of the bright and dark areas effectively;Then,we employ the non-linear brightness enhancement method and the contrast enhancement method based on neighborhood correlation to enhance the two regions respectively,so that the bright and dark regions have higher brightness and contrast values.Finally,to complete the entire process of enhancement,we apply the fusion method based on the Laplacian pyramid to merge high-quality bright and dark areas.Aiming at this method,this paper compares the experiments with the existing six enhancement methods,and analyzes the experimental results through the visual effects of images and objective evaluation indicators.The results show that the method proposed in this paper achieves better results under both subjective visual effects and objective evaluation indicators:the brightness of the dark area is greatly improved under the condition of less noise,so that the enhanced result has a better visual effect.Aiming at the above-mentioned algorithm's poor recovery of the color and detail information of the background area,a multi-scale structure fusion algorithm based on secondary illumination estimation is proposed.First,we propose a secondary illumination estimation method.The first illumination estimation method is used to perform effective enhancement processing on the backlit area,and the second illumination estimation method is used to restore the color and detailed information of the bright area to better distinguish bright and dark areas.Then,we use the multi-scale fusion method based on the structure decomposition model to fuse the well-exposed parts of bright and dark areas to obtain the final result.Aiming at this method,this paper compares the experiments with the existing five enhancement methods,and analyzes the experimental results through the visual effects of the image and objective evaluation indicators.The results show that the method in this paper has achieved better results:the algorithm recovers more color information and detailed information on the basis of the first algorithm,and also has a higher contrast.