Research On The Near Infrared Vein Image Recognition And Processing Algorithms

Venipuncture is one of the most common technology in modern medical domain,appied in medical industry,including intravenous injection,intravenous fusion,venous blood collection and transfusion,and so on.In real life,the vein can be observed in relatively is the primary condition of venipuncture that was vulnerable to the skin color,the deepness of vein and obese condition,which would lead the venipuncture mistake and depend on the experience of the medical staff.Thus,it is significant practical application value to study the recognition and processing algorithm of vein image to help venipuncture.Making use of the sensitiveness of near-infrared light,the general vein image was informed.The challenge of the original image,such as the inhomogeneous illumination,the low contrast and the unclear edge would augment the difficulty in venipuncture.This paper has proposed a way to promote the vein image by improving the device of sampling near-infrared vein,then use the improved arithmetic to deal with the vein image,and put forward a measurement method of hand vein subcutaneous.For starters,the work has improved the non-invasive imaging of vein based on FPGA utilizing the theory that the vein and the organizations around have different absorption and reflection of near-infrared.And the author has designed and set up the mounting bracket,modified the double layer acrylic optical diffusion plate,added an optical diffusion near infrared filter,and through the survey of vein image acquisition conditions,transmitted the image to computer system,and then deal with the image in time.In image processing,firstly,this paper propose an algorithm to automatically extract the region of interest which can be easy to make vein image filtering denoising,image enhancing and edge detection.Second,the paper has proposed the improved algorithm─restricted median filter denoising algorithm,histogram equalization contrast enhancement algorithm based on gamma transformation and edge detection algorithm based on Otsu threshold segmentation for filtering the noise of veinimage,improving the color of image,and testing the edge.The improved algorithm is simple and flexible,what is more,the results are as the same as the predicted one.Thirdly,the paper has evaluated the improved algorithm,and evaluation results show that the improved algorithm has certain advantages compared with the traditional algorithm,and the improved algorithm is simple and flexible.At last this work proposed gray correspondence algorism to measure the deepness of human hand vein.The process used a man-made model of a dorsal hand then calibrated the gray value of model image with the deepness of hand model vein and make curve fitting,which can measure the depth of human dorsal hand vein,give the visual difference and spot the depth in time in order to help the medical staff estimate the condition of venipuncture and achieve auxiliary applications of medical.