Vessel injections are a group of injections which drug material is directly injected into the body veins. Since superficial veins are not easily seen in some individuals, these injections encounter many difficulties and injections have to be done by try and error in these situations. Because of that, a new system is designed and implemented in this thesis to assist the physicians in these injections. This system applies image processing techniques to show hidden body veins directly on the body skin. The main idea is to capture vein images in near infrared band, because veins are more apparent in this band. Then the necessary transformations and enhancements are applied to these images and finally the resulted images are projected back on the captured area by a projector. In this manner, deep veins would be visible to human naked eye directly on the body skin. The projected image should be aligned precisely with the real body veins. So it is necessary to calibrate the camera and the projector. In a former system, high precision micrometer screw adjusted mechanical stages were used to set the optimal camera-projector positions and the transformation function between the captured image and projected image was estimated using a simple affine transform (by four circular control points located on the rectangular vertices). In this thesis, such mechanical parts are not used. To achieve the needed alignment, a complicated model is used to transform the captured image into the projected image. This transformation is estimated using the following procedure. A cheoard image is projected on the target by the projector and an image is captured using the camera. The transformation between these two images is estimated using a group of control points. These control points are located on the vertices of the cheoard image and should be extracted automatically. This is not a trivial task because of some bothersome factors (hairs, blotch, …), noise, low contrast and nonuniform illumination. So at the first step, these bothersome factors should be eliminated in the captured images and in the next step, all corners should be extracted. We used two robust methods to extract corners robustly. The success rate of our method for corner detection is 100%. After corner detection, the transformation function between the camera and projector is estimated using the extracted corners. Global nonplanar or local planar models are a better choice in this application because of the curvature of the body. Different transformations are tested and the best is selected by considering both the accuracy and the run time. Forth order polynomial is shown to be the optimal model in this respect. Unsharp Masking and contrast-limited adaptive histogram equalization are applied to enhance image quality. Different high contrast color schemes can also be applied to these images for better visualization. Finally the transformed and enhanced images are projected back on the desired area with the accuracy of 0.089 mm using a Pico-projector. The applied Pico-projector is in the pocket dimensions and uses DLP technology to make images on the target area. Keywords: Vein contrast enhancer, vein viewer, vein finder, camera-projector calibration, spatial transformations, contrast enhancement