论文部分内容阅读
目的 :提出一种基于特征点位置坐标的定量描述颅面x光片图像的相似性的方法 ,并应用到研究颅面X光片图像正常值的确定中。方法 :提取两幅X线头影测量图像中的特征点的位置坐标 ,构成两组二维向量 ,用典型相关分析这两组向量的相关性 ,该相关性就表征了两幅图像间的相似度。根据图像的相似性 ,获得各个图像叠加时的权重 ,从而确定正常值。结果 :用该方法分析 1 0例男性恒牙期正常颅面X光片 ,得到每两幅图之间的相似度和各图叠加时的加权系数。并对影响相似度的因素作了分析 ,说明特征点的数目和选取方式对相似度有影响 ,相似度越低 ,越容易受影响。重复定点中的随机误差对相似度的影响很小。结论 :该方法能够定量地描述图像的相似度 ,对研究颌面正常值具有一定的意义 ,并可以直接推广到研究 3D医学图像的相似度和正常值
OBJECTIVE: To propose a method for quantitatively describing the similarity of craniofacial images based on the location coordinates of feature points, and to apply it to the determination of the normal value of craniofacial X-ray images. Methods: The position coordinates of feature points in two X-cephalograms were extracted to form two groups of two-dimensional vectors. Canonical correlation analysis was used to analyze the correlation between the two groups of vectors. The correlation represented the similarity between the two images . According to the similarity of the images, the weight of each image when superimposed is obtained, so as to determine the normal value. Results: The method was used to analyze 10 normal permanent craniofacial X-ray films of permanent men. The similarity between every two images and the weighted coefficients of each figure were obtained. And the factors that affect the similarity are analyzed. It shows that the number of feature points and the selection method have an impact on the similarity. The lower the similarity is, the easier it is to be affected. The effect of random errors in repetitive fixed points on the similarity is small. Conclusion: This method can quantitatively describe the similarity of images and has certain significance for the study of normal value of maxillofacial region. It can be directly extended to study the similarity and normal value of 3D medical images