A Robust Fiducial Mark Extraction Method in X-ray Image based on HOG Operator

摘要

Automatic extraction of fiducial mark in X-ray images[1～2] is an important research field in surgical navigation systems. Traditional Hough circle detection and edge detection are lack of accuracy. This paper propose a robust extraction method for fiducial mark in X-ray images based on the Histogram of Oriented Gradients (HOG)[3] operator. The process of fiducial mark extraction has two steps: the first step is to do the coarse match based on the gray distribution of the Normalized Cross Correlation (NCC)[4] method; the second step is to calculate HOG features of the candidate points and classify the features by the Support Vector Machine (SVM)[5]. We use the Ray-casting[6] method to get the multiple matching templates. The experimental results show that the proposed algorithm can achieve automatic detection of marker points in X-ray images, and has high accuracy and robustness.