The Research And Comparison On Characteristic Extraction Method For 2-D Skull Based On CT Image

Biological recognition technology, as a hotspot of research and application, hasdeveloped quickly in last few years. The identification technology depending on biometricsuch as human face, fingerprint, iris etc, are growing maturity and applied broadly in manyfields. Skull recognition, as a kind of biological recognition technology, hasn't got enoughattentions, but considering biological recognition technology's developments and the intrinsicfeatures of skull recognition, the developing chance of skull recognition is closing near. Themain target of this research is skull image. Characteristic of skull will be extracted as a basisfor the identification of different individuals. For the purpose, we have an in-depth study forkey technologies and major algorithms in identification process. The main work includes thefollowing:(1) On the basis of analysis and review in the relevant research result and latest researchprogress, this paper has sum up the purpose and signification of skull study.(2) A few common algorithms about boundary description are used in skull recognition.We analyze advantages and disadvantages of these algorithms applied in skull recognitionthrough a lot of experiments.(3) A new algorithm about skull boundary recognition is provided, which is Legenderdescription of polar radius mean difference. The validity of the algorithm in skull recognitionis proved by experiments.(4) Combining function of center projection mean difference and Chebyshev orthogonalpolynomials, a new description of the gray image is attained. Experiments show that thealgorithm is effective in skull recognition.(5)According to the research of local skull information, we try to recognize skull by usingtemplate matching algorithm and skull feature-point matching algorithm.In this process, we focus on the research of identification algorithm. On the basis ofcomparing and analyzing the existed algorithms, some novel algorithms are brought forward.A lot of experiments show that skull recognition is absolutely doable and can achieve goodresults.