A new approach aimed at the registration of three-dimensional(3-D) coronarographic images is reported. The matching of 3-D vessel central lines is performed using attributed string matching techniques. Firstly

we extend the 2-D Freeman chain code to 3-D space. We then give a brief description of string matching techniques

with a special emphasis on image analysis. Some key elements involved in the algorithm such as chain encoding

cost function

normalized edit distance are proposed and discussed. In this paper

two types of substitution cost functions are introduced and used in string matching algorithm. Their performances are compared on experimental data obtained and further evaluate the results by means of subjective and objective criteria. The experimental results show that the use of the second substitution cost function can yield better matching results.