朱望 沈疆海
摘要:巖心分析在地質(zhì)巖心研究中具有重要作用,是科技人員了解地下信息的重要依據(jù)。利用巖心掃描儀將巖心高精度掃描下來(lái)得到的巖心圖像,是研究人員進(jìn)行巖性分析的重要材料。但是掃描攝像頭視域有限,只能采取分段掃描再拼接的方法來(lái)構(gòu)成一幅完整的巖心圖像。為了滿足這一需求,本文利用SURF實(shí)現(xiàn)了圖像特征點(diǎn)的快速匹配,利用FLANN+Knnmatch近鄰值算法降低了拼接對(duì)圖像重疊率的要求,利用RHO算法提升拼接成功率,最后利用漸入漸出加權(quán)平均法融合拼接縫,完成巖心圖像的無(wú)縫拼接。相對(duì)于傳統(tǒng)算法,本算法拼接速度快、對(duì)圖像質(zhì)量要求更低、拼接成功率高,并能完美融合拼接縫,達(dá)到很好的實(shí)用效果。
關(guān)鍵詞:巖心圖像拼接;SURF;無(wú)縫拼接;圖像融合
中圖分類(lèi)號(hào):TP311? ? ?文獻(xiàn)標(biāo)識(shí)碼:A
文章編號(hào):1009-3044(2021)26-0105-03
開(kāi)放科學(xué)(資源服務(wù))標(biāo)識(shí)碼(OSID):
An Improved Algorithm for Seamless Stitching of Core Images
ZHU Wang, SHEN Jiang-hai
(School of Computer Science, Yangtze University, Jingzhou 434023, China)
Abstract: Core analysis plays an important role in geological core research, and is an important basis for scientific and technological personnel to understand underground information. The core image obtained by high-precision core scanning by core scanner is an important material for the researchers to conduct lithology analysis. But the visual threshold of the scanning camera is limited, so the method of segmented scanning and splicing can only be used to form a complete core image. In order to achieve this demand, In this paper, surf is used to realize the fast matching of image feature points, FLANN + knmatch algorithm is used to reduce the requirement of image overlap rate, Rho algorithm is used to improve the success rate of stitching, and finally the gradual in gradual out weighted average method is used to fuse the stitching seam to complete the seamless stitching of core image. Compared with the traditional algorithm, this algorithm has the advantages of fast stitching speed, lower image quality requirements, high stitching success rate, perfect fusion of stitching seams, and good practical effect.
Key words: core image stitching; SURF; seamless stitching; image fusion
1 研究背景介紹
在地質(zhì)巖心研究中,對(duì)巖心圖像進(jìn)行分析是了解地下信息的重要途徑。高分辨率巖心圖像主要通過(guò)巖心掃描儀掃描得到。開(kāi)始掃描時(shí),攝像頭下方的兩個(gè)水平滾動(dòng)軸沿同一方向緩慢自傳,帶動(dòng)滾動(dòng)軸上的圓柱狀巖心轉(zhuǎn)動(dòng),同時(shí)上方攝像頭對(duì)巖心進(jìn)行持續(xù)性掃描。因?yàn)閿z像頭視域問(wèn)題,每對(duì)一個(gè)視域內(nèi)的巖心完成圓周掃描,攝像頭就會(huì)水平滑動(dòng)至下一個(gè)視域,繼續(xù)進(jìn)行一個(gè)圓周掃描,如此往復(fù),直至掃描完整段巖心。掃描結(jié)束后,電腦獲得的是一張張不連續(xù)的巖心圖像,為了支撐研究人員對(duì)地質(zhì)巖心進(jìn)行宏觀分析,需要將不連續(xù)的圖像拼接成整幅完整的巖心圖像。為了解決這個(gè)問(wèn)題,巖心圖像拼接是一項(xiàng)必要工作。
目前對(duì)于巖心圖像拼接算法的研究都已經(jīng)取得了相對(duì)理想的效果。Zhang Xiao等利用改進(jìn)的ORB算法提取特征點(diǎn),采用多分辨率融合算法對(duì)圖像進(jìn)行平滑拼接[2]。Liu Yue等將圖像按頻率進(jìn)行分解,分別融合各個(gè)頻率上的帶通圖像,再反變換還原出融合結(jié)果,實(shí)現(xiàn)巖心圖像無(wú)縫拼接[3]。Liang Tao等利用改進(jìn)的Harris角點(diǎn)算法對(duì)巖心圖像進(jìn)行檢測(cè),為每個(gè)角點(diǎn)生成特征描述,最后使用馬氏距離進(jìn)行加權(quán)平均和特征匹配[4]。Wu Xiaohong等利用SIFT算法來(lái)提取巖心圖像的特征點(diǎn),采用歐氏距離進(jìn)行特征匹配,自動(dòng)配準(zhǔn)高清巖心圖像[5]。但是由于巖心圖像的尺寸較大、分辨率較高的原因,以上算法在拼接速度、成功率與實(shí)用性上無(wú)法兼得。為此,本文利用改進(jìn)的SURF算法進(jìn)行圖像配準(zhǔn),漸入漸出加權(quán)平均法實(shí)現(xiàn)拼接縫融合,本算法在保證拼接速度的同時(shí),極大地提升了拼接成功率,針對(duì)巖心掃描及時(shí)拼接有很好的實(shí)用性。