The Registration Algorithm Based On CPD In Application Of Femoral3D Reconstruction

The3D reconstruction of femur according to the data of the CT, DR images, and ICP algorithm as the foundation, realizes the visualization of3D posture,this technology has a lot of advantages such as a low cost, a small radiation to human body and also has a higher efficiency in processing image, it can provideconvenience image data for medical theory and practice research. The imageregistration algorithm is the main research content of the femur3Dreconstruction, registration parameters were obtained through the auxiliarymachine determines the accuracy of femur3D reconstruction. The ICP is acommon registration algorithm, because the algorithm has the disadvantages ofhigher complexity and robustness, as while as low accuracy, this paper proposesa registration algorithm based on CPD, makes registration process more accurateand automated.First of all, this paper elaborates the important principle of imageregistration, including the similarity measure, optimization method, and thespace transformation; Then according to the transform method, imagedimensions and image characteristics to summarize the registration methods indetail; And introduced the key steps and evaluation standard of imageregistration in a simple way, do set the stage for the following research work ofthis paper.Secondly, this paper introduces the basic theory and application field of thetraditional ICP algorithm, aim at the low accuracy and robustness of thealgorithm, and then puts forward important principle and research content of theregistration algorithm based on CPD. The algorithm not only considering thediversity and similarity of general femur and patient femur, but also joined theproportional deformation, success to avoid the disappoingtment of thetraditional ICP algorithm, is superior to the traditional ICP in time and space complexity, make registration process more accurate. Then propose theexperiment process of registration algorithm based on CPD mainly, mainlyincluding the preprocessing of image, the acquisition of registration data, andthe registration of CPD points.Finally, through a lot of simulation evaluation of femur and the verificationalgorithm of this paper, the experimental results are as follows: the average errorof translation and rotation is less than2.1mm,1.6°respectively; Registrationfailure rate is only around10%; the similarity of3D model from patient femoralis much higher. The results show that the proposed registration algorithm basedon CPD in this paper has a good robustness, high precision, lower complexityand a better registration effect.