Contour pick-up and registration of infrared images of palm

In this thesis an approach for contour pick-up and image registration is presented for infrared images of the palm. Contour pick-up is based on Gauss-Laplace kernel and image registration is implemented by their cross-correlation function. In order to use the temperature variation obtained from the infrared images for diagnoses of metabolic diseases, one used to have to obtain the temperatures at different times for a certain area artificially point by point. It always wastes time and the data cannot be very accurate. This thesis gives a method that avoids these disadvantages. First, the contour is picked up from the environment by using the convolution of the Gauss-Laplace kernel. Then we filled inside the edge with darkness. Counting the maximum value of their cross-correlation function realized registration of the infrared images of the palm. The results of experiments prove this method to be feasible.