本文選題：醫(yī)學圖像匹配 + 移動隊列規(guī)則　； 參考：《西南大學學報(自然科學版)》2017年12期

【摘要】：當前醫(yī)學圖像的特征匹配主要依靠像素灰度來完成,但是像素灰度對空間信息不敏感,當匹配圖像之間存在灰度信息不均衡以及噪聲干擾時,將導致誤匹配率較高,對此,本文提出了一種基于移動隊列規(guī)則耦合角度約束的醫(yī)學圖像匹配算法.首先,利用高斯金字塔模型對源圖像進行濾波預(yù)處理,以減少源圖像中存在的噪聲等干擾;再利用Harris算子對預(yù)處理后的源圖像進行特征檢測,獲取圖像的特征點;然后,利用SURF(Speed Up Robust Feature)特征描述子,獲取特征點對應(yīng)的特征描述子.并通過尺度空間理論獲取特征點集,通過將特征點集進行排序來形成隊列,從而設(shè)計移動隊列規(guī)則,完成特征點的匹配;最后,通過求取匹配特征點間的夾角,形成角度約束模型,對匹配特征點進行提純,剔除偽匹配特征點,使得匹配準確度得以提升.從仿真實驗結(jié)果與分析可見,在對醫(yī)學圖像進行匹配時,本文所提出的方法具有匹配精度高、魯棒性能好等特點.
[Abstract]:At present, the feature matching of medical image mainly depends on pixel gray level, but pixel gray level is not sensitive to spatial information. When there is imbalance of gray level information and noise interference between matching images, the mismatch rate will be high. In this paper, a medical image matching algorithm based on the coupling angle constraint of mobile queue rules is proposed. Firstly, using Gao Si pyramid model to filter and preprocess the source image to reduce the noise and other interference in the source image, then using the Harris operator to detect the feature of the preprocessed source image and obtain the feature points of the image. The feature descriptor corresponding to feature points is obtained by using SURF(Speed up Robust feature descriptor. And through the theory of scale space to obtain the feature points set, by sorting the feature points set to form a queue, so as to design the mobile queue rules, complete the matching of feature points; finally, through the calculation of the matching angle between feature points, The angle constraint model is formed, the matching feature points are purified, and the pseudo matching feature points are eliminated, so the matching accuracy can be improved. From the simulation results and analysis, it can be seen that the method proposed in this paper has the characteristics of high matching accuracy and good robustness when matching medical images.
【作者單位】： 江蘇醫(yī)藥職業(yè)學院圖書信息中心;
【基金】：江蘇省自然科學基金項目(BK2015609)
【分類號】：R318;TP391.41
，

本文編號：1808063