| Personal identification technology is applied to a wide range of life. It makes use of the information of people's biologic characteristic to recognize person's identity. Finger-vein recognition is one of such technology. In this system, an infrared light is transmitted from the backside of the hand. A finger is placed between the infrared light source and camera, As hemoglobin in the blood absorbs the infrared light, the pattern of veins in the palm side of the hand is captured as a pattern of shadows. The captured images contain not only vein patterns but also irregular shading and noise, the shading is produced by the varying thickness of finger bones and muscles. Therefore, regions in which the veins are and are not sharply visible exist in a single image. To solve this problem, we need to preprocess the finger-vein image, and finally obtain clear pattern of the finger-vein. This is our research background.In this paper, we propose a new method to enhance the contrast of the finger-vein image,. It consists of three parts: denoising,contrast enhancement and binary image. We will introduce them separately(一) .DenoisingThe purpose of denoising is to reduce or eliminate the white noise in the vein image, and avoid affecting the enhancement effect. In this paper, we use wavelet-transform method.As we know, Using DWT (Discrete Wavelet Transform), we can transform the signal into its different frequency bands, and for each band, we design different method of denoising .First we perform stationary wavelet decomposition, and transform the image into four frequency bands: low frequency LL, middle frequency LH and HL, High frequency HH. For low frequency LL, we use soft-threshold method to maintain the smoothness of the image. For middle frequency LH and HL, because LH and HL contain the edge information of the vein-image, and most of them are local extrema. So we only save these extrema of the middle frequency band, and other wavelet coefficients are set to be zero. For high frequency HH, because it consists of isolate white noise, so we simply apply thresholding method to eliminate noise. Finally we do stationary wavelet reconstruction ,and obtain the denoising image.(二) Contrast EnhancementTraditional whole image histogram equalization method obtains good result in contrast enhancement of common image. But when applying this method to finger-vein image, we meet with a problem: during the Acquisition of the vein image, due to the varying thickness of the finger bones and muscles,light, etc, the image obtained has some regions which are entirely dark or bright, if we apply histogram equalization method to the whole image, the regions which are dark will become darker, and the regions which are bright will become brighter. Therefore, the patterns of the vein are hidden.One of the methods to solve this problem is to divide the image into several parts, and apply histogram equalization to each part . The result shows that the pattern of the finger-vein is clearly revealed in each part, but unfortunately, we also introduce block noise into the image. Even worse ,on the edge of each part, we introduce false vein pattern into the image , they are hard to eliminate . So dividing the image is not a property method.The other method to solve this problem is the method we proposed in this paper: histogram template equalization: First, we select a threshold to separate the finger from the background .This threshold is easy to select , because the finger is sharply distinguished from the background.Then we create a template, the size of which is set to be M×N,(here, M and N are both odd numbers).The center of the template is C (N/2, M/2). Select a pixel ( x0 , y 0), which is in the finger, as the start pixel. Suppose it's gray value is f ( x0 , y 0).Then put the template on the vein image ,such that the center of the template is overlapped with ( x0 , y 0) .So there are Q=M×N pixels in the template(pixels which are not in the finger is not considered, in this case Q |
PDF Full Text Request