CAJ | 학술논문

针对肖像类唐卡图像头光区域呈圆形的特点，将圆检测算法应用到肖像类唐卡图像头光区域的定位中，并在此基础之上提出了一种自动分割头饰的算法。该算法首先利用圆检测算法定位出头光所在的大概位置；然后利用头光颜色的空间分布特征、头饰外轮廓以及边缘检测结果实现头饰分割。实验结果表明，该算法在肖像类唐卡图像头饰的分割中具有头饰定位准确的特点，并且主尊比例和头饰大小的变化不会影响分割效果。
침대초상류당잡도상두광구역정원형적특점，장원검측산법응용도초상류당잡도상두광구역적정위중，병재차기출지상제출료일충자동분할두식적산법。해산법수선이용원검측산법정위출두광소재적대개위치；연후이용두광안색적공간분포특정、두식외륜곽이급변연검측결과실현두식분할。실험결과표명，해산법재초상류당잡도상두식적분할중구유두식정위준학적특점，병차주존비례화두식대소적변화불회영향분할효과。
Aiming at the characteristic of circular back-light of headdress in Thangka portraits,circle detection algorithm is applied to back-light region location of headdress in this article,and a new headdress segmentation algorithm is proposed basing on the region location method. The algorithm firstly uses the circle detection algorithm to locate the general location of back-light of headdress;then it uses the color spatial distribution of back-light of headdress,outer contour of headdress and the result of edge detection to segment headdress. Experiments show that the algorithm in this article is characterized by accurate location of headdress,and the proportion of main statue and the size of headdress change do not affect the segmentation effect.