基于形態(tài)學(xué)的分水嶺算法牙科x射線圖像分割-資料下載頁
2025-06-27 20:42本頁面
【正文】 riginal dental Xray image. And (b) shows the result after enhancement. Apparently, the objectives (teeth) and details are highlighted obviously and the contrast of foreground and background (soft tissues) are amplified. And we get the binary image of dental Xray image in (c). And it is evident that adhesion between each tooth is weakened after the first erosion in (d). The structure element b of this processing cannot be too big for the chance of oversegment happens. (e) shows the result of applying the function of imfill() to the image. And some holes are filled to avoid the phenomenon of oversegment caused by the second erosion. And after two erosions the adhesion are clearly eliminated as showed in (f). The spilt lines we get after applying the proposed algorithm shows in (g). From (h) that is the original image with the segment line and we can see that this algorithm can satisfactory, successfully segment each tooth with oversegmentation effectively improved.The image applied with classical watershed algorithm using distance transform is presented in Figure 4. Compared with Figure 3(c) we can see that part of those teeth cannot showed in Figure 4(a) without enhancement and this could lead to the incorrect split lines such as the one of the middle tooth. And From Figure 4(b) and (c) we can see that we cannot get the proper contour information just applying the classical watershed algorithm because the touching teeth belong to one area not isolate areas with their adhesion.Figure 5 and figure 6 show another two examples used the proposed algorithm. From the results showed in these figures we can see that the of upper jaw or to the lower jaw are segmented precisely and accurately. But the lines we have got splitting each tooth still need to be improved. And this phenomenon like Figure 5(b) shows can be explained by the structure elements of b used to delete the adhesion between teeth. The erosion cannot avoid to the reduction of the teeth areas cannot be avoided by implementing the operation of erosion. IV. CONCLUSIONS AND FUTURE WORKFor the purpose of separate each tooth, this paper presents an algorithm of watershed transformation based on morphology and distance transform. According to the results of the experiments, this algorithm can improve the phenomenon of oversegmentation and split each tooth effectively. But the results of segmentation are not perfect and it takes time to adjust the parameters like the structure elements b .Our future work will concentrate on choosing an auto adaptive capability of the structural elements. And we also intend to apply the results of segmentation to more extensive and deeper areas like human identificat
研究報(bào)告相關(guān)推薦
鄂ICP備17016276號(hào)-1