Design of An Automated Injection System Based on Dorsal Hand Vessel Recognition

Dorsal hand venipuncture is one of the most common tools used in clinical care, but there are fewer studies addressing the identification of dorsal hand veins. In real life, venipuncture often fails due to various reasons such as skin color or obesity. Therefore, it is of great practical importance to design a system with clear NIR vascular visualization and automatic injection. This thesis describes a system for identifying and localizing the veins on the back of the hand for detection. The global dorsal hand vein images were acquired using the MT9V034 CMOS camera under S50 nm NIR light irradiation. Based on this, a threshold value is acquired by the Otsu method to remove the background part outside the back of the hand, and then the vascular image is enhanced by a data enhancement algorithm to obtain a clear vascular image. Then, the blood vessels are segmented using segmentation algorithm to achieve precise positioning of blood vessels, and finally, automatic injection is performed by robotic arm. The experimental results show that the method can achieve the identification of blood vessels and the operation of their automatic injection.