CAJ | 학술논문

引入图像处理技术进行骨龄识别是目前骨龄研究的热点，提出一种基于k-余弦曲率和小波支持向量机（ WSVM ）的骨龄识别方法。首先对手腕骨X光图像进行预处理，通过k-余弦曲率算法定位指骨关键点，然后分割指骨特征区域，提取骨骺发育过程中的几何特征作为骨龄的研究参数，最后，根据“中国人手腕骨发育标准———CHN法”，利用WSVM多分类算法对特征骨块按等级进行分类，综合各个骨骼的等级识别骨龄。实验表明，该方法能有效地对特征骨块进行分期，从而对骨龄识别具有较高的识别率和稳定性。
인입도상처리기술진행골령식별시목전골령연구적열점，제출일충기우k-여현곡솔화소파지지향량궤（ WSVM ）적골령식별방법。수선대수완골X광도상진행예처리，통과k-여현곡솔산법정위지골관건점，연후분할지골특정구역，제취골후발육과정중적궤하특정작위골령적연구삼수，최후，근거“중국인수완골발육표준———CHN법”，이용WSVM다분류산법대특정골괴안등급진행분류，종합각개골격적등급식별골령。실험표명，해방법능유효지대특정골괴진행분기，종이대골령식별구유교고적식별솔화은정성。
To recognise the bone age by introducing image processing technology becomes the focus of current bone age research.In this paper we propose a bone age recognition method, which is based on k-cosine curvature and wavelet support vector machine ( WSVM) .First we preprocess the X-ray images of carpal, and locate the key points of phalanges by k-cosine curvature algorithm.Then we segment the phalanges feature region EMROI, and extract the geometric features characterising the growth of epiphyses as the research parameters of bone age.In the end, we use WSVM multiclass classification algorithm to classify the feature bone blocks in different grades according to the“Standard of Chinese Carpal Development-CHN, and recognise the bone age by integrating the grades of different bones.Experiments show that the method can effectively sort the feature bones in stages so that it has higher recognition accuracy and stability on bone age assessment.