Research On Feature Detection Algorithm Of X-ray Imaging Mark Point For Medical Image Registration

In the surgical navigation systems,the fiducial markers that embedded in human body are used to assist image registration of X-ray images and CT images.However,the current technology,if you want to determine the location of the mark points needs to experts manual identify and practical experience.Automatic extraction of X-ray image marks is significant for the full automation of surgical navigation systems.This paper is based on an actual research project.The purpose of this paper is to study and implement the feature detection algorithm of X-ray image mark points in medical image registration,including the construction of mark point template,the research and Innovation of feature description algorithms,and the detection of X-ray image mark points with SVM learning classification algorithm.The specific content of this work and research results are as follows:(1)Research on the methods of constructing X-image marker point template.Firstly,we analyze the image features of mark points with different imaging sites and different imaging angles.Then we study two kinds of traditional mark point template construction methods,and propose a method of marking point template construction based on Ray-casting algorithm.Experiments show that the use of the template construction method proposed in this paper can improve the detection accuracy of the mark points.(2)Research on an improved algorithm of the marker point feature detection.Based on the coarse and fine matching theory,the rough matching and rough matching results of the NCC algorithm are used as the candidate points of the matching points,and the matching process based on the feature matching algorithm is studied and realized.In the process of fine matching process,the feature description algorithm for HOG is proposed to solve the problem of point distortion,background overlap and edge blur.Including HOG improvement of Marker Gradient Gaussian Weighting and HOG of Image Blocking Method,and using the integral gradient histogram to speed up the algorithm processing.The experimental results show that the improved algorithm of mark point feature detection can improve the robustness of mark detection.(3)Research of the X-ray image marker points detection based on SVM.The positive and negative samples of the marker image are obtained by the NCC rough match.Then,the HOG algorithm proposed in this paper is used to extract features from the positive and negative samples and use the features as the input of the SVM.And then,the trained SVM classification model is used to realize the detection of marker points.The experimental results show that the improved algorithm of feature detection proposed in this paper can realize the detection of marker points in X-ray image with the help of SVM classifier.Finally,the points that are detected are used for medical image registration.The innovation work of this paper mainly includes two points:(1)Aiming at the problem of template construction in marker point detection,the Ray-casting algorithm is applied to the construction of marker p6 int template,which improves the detection performance of marker point.(2)Aiming at the marker point,a new algorithm for feature description based on HOG algorithm is proposed.