The Study On Multi-Exposure Image Fusion Of Finger Vein

In recent years,with the rapid development of the Internet,Internet information has also increased exponentially.Therefore,the issue of information security has received extensive attention from all walks of life.The finger vein recognition technology has attracted the attention of the society and academic research by virtue of its uniqueness,high precision and high anti-counterfeiting.However,there are still deficiencies in the process of finger vein image acquisition at this stage.Because the finger vein belongs to the internal characteristics of the human body,and because the collection device is a traditional CMOS image sensor,and the collection environment is a semi-closed structure,the image details of the finger vein collection are relatively missing,and the brightness contrast is not very obvious,thus affecting The subsequent process of finger vein recognition(such as feature extraction,feature matching,etc.)is performed.In order to obtain this high-dynamic range finger vein image,this paper has carried out research based on high dynamic range imaging technology,which has high research and implementation value.This paper presents a multi-exposure image fusion algorithm based on camera response curve and Laplacian pyramid.Firstly,the finger vein multiple exposure image is obtained by the finger vein image acquisition device,the camera response curve is obtained by the pixel value and the exposure time corresponding to each pixel of the multi-exposure image,and then the information amount of each pixel point is determined according to the camera response curve.And the Gaussian equation constrains the brightness of the image and assigns weights to it.At the same time,a weight correction function is proposed to avoid the loss of image details in overexposed or underexposed regions,and the final weight function is obtained.Then,the Gaussian pyramid decomposition of the multi-exposure image and the weight function is performed,and the image fusion is performed at multiple resolutions,and the weight factors of different levels are combined with the image details at different levels,so that the details of the small range can be grasped,and the details can be grasped.The overall characteristics of the image.Finally,the image is reconstructed and synthesized based on the Laplacian pyramid to obtain the final finger vein multi-exposure fusion image.Compared with the previous multi-exposure image fusion algorithm,the algorithm is simple and feasible.The image details of over-exposed and under-exposed areas in the finger vein image are preserved as much as possible.The vein contrast is obvious,the branch extension is more clearly visible,and the color is reduced.Distortion provides a good image basis for finger vein recognition.A multi-exposure fusion image ghost elimination algorithm based on luminance value relationship in dynamic scene is proposed.It is assumed that N images of different exposures taken by the same shooting device for the same scene are input.Then,an optimal reference image is selected by the algorithm,and the brightness of other images is unified according to the reference image brightness level.Then use the difference method to calculate the difference between the pixel values between the reference image and other images,and compare the difference with the threshold set herein.If it is larger than the area,it is the motion area,then the pixels at the position are removed.,that is,the weight is set to 0;otherwise,it is considered to be reserved by the still region,and the weight is set to 1.Finally,the binary image function is added to the weight function of the multi-exposure image fusion to obtain a new Gaussian pyramid weight map,and then Laplacian pyramid decomposition,fusion and reconstruction are performed for each exposure image,and ghost ghost elimination is finally obtained.The subsequent finger vein multiple exposure fusion image.The algorithm can not only select the optimal reference image through the algorithm,but also select the reference image manually,which is convenient for the user to get the final fusion image of the ghost image that he wants.The biggest advantage of the proposed algorithm is that even if there are a large number of overexposed or underexposed areas in the reference image,dynamic ghosting can be well eliminated without image distortion and halo phenomenon.