《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 其他 > 設(shè)計應(yīng)用 > 基于熵率聚類的超像素機器視覺與缺陷檢測算法
基于熵率聚類的超像素機器視覺與缺陷檢測算法
信息技術(shù)與網(wǎng)絡(luò)安全
李 鋒
(廣東交通職業(yè)技術(shù)學院,廣東 廣州510650)
摘要: 在智能制造中,傳統(tǒng)成像技術(shù)已經(jīng)滿足不了高精度工業(yè)需求。提出了結(jié)合熵率聚類的目標分割算法,并且基于超像素的鄰邊集,建立熵率和平衡項的目標函數(shù),最后通過貪婪啟發(fā)算法優(yōu)化并求解該目標函數(shù),得到最優(yōu)的超像素集合。并設(shè)計了基于高斯函數(shù)衡量相鄰像素的相似性實驗,設(shè)定相關(guān)參數(shù),進行工業(yè)制造實際流程檢測。最終實驗結(jié)果表明,所提算法有較好的檢測識別效果,在輪廓及內(nèi)部條紋識別上效果明顯,整體識別效果良好,適用于工業(yè)制造領(lǐng)域。
中圖分類號: TP393
文獻標識碼: A
DOI: 10.19358/j.issn.2096-5133.2021.02.012
引用格式: 李鋒. 基于熵率聚類的超像素機器視覺與缺陷檢測算法[J].信息技術(shù)與網(wǎng)絡(luò)安全,2021,40(2):70-73.
Super pixel machine vision and defect detection algorithm based on entropy rate clustering
Li Feng
(Guangdong Communication Polytechnic,Guangzhou 510650,China)
Abstract: In intelligent manufacturing, traditional imaging technology can no longer meet the needs of high-precision industry. In this paper, a target segmentation algorithm combining entropy rate clustering was proposed, and the objective function of entropy rate and equilibrium term was established based on the adjacent edge set of hyper pixel. Finally, the optimal hyper pixel set was obtained by optimizing and solving the objective function through greedy heuristic algorithm. A similarity experiment based on Gaussian function was designed to measure the similarity of adjacent pixels, and the relevant parameters were set to test the actual process of industrial manufacturing. The final experimental result shows that the algorithm has a good detection and recognition effect, is obvious in contour and internal fringe recognition, and the overall result is good, which is applicable to the field of industrial manufacturing.
Key words : machine vision;entropy clustering;super pixel;greedy heuristic algorithm

0 引言

         隨著智能制造工藝精度的提高,高精度和快速檢測成為目前亟待解決的問題。機器視覺與圖像識別作為非接觸式檢測方式,具有檢測速度快、精度高的特點,能很好地解決智能制造流水線中的瓶頸,并逐步替代傳統(tǒng)人工檢測方法。

        工業(yè)檢測對表面缺陷檢測要求更嚴格,傳統(tǒng)表面缺陷成像方法,包括線掃描、結(jié)構(gòu)光、面陣相機等已經(jīng)不能滿足精度要求,基于超像素檢測算法由此誕生。表面缺陷檢測問題包括圖像分類和圖像分割兩大部分,通過采集大量缺陷與合格產(chǎn)品圖像,對比分析圖像中缺陷特征,設(shè)計相應(yīng)缺陷檢測算法。




本文詳細內(nèi)容請下載:http://ihrv.cn/resource/share/2000003382




作者信息:

李  鋒

(廣東交通職業(yè)技術(shù)學院,廣東 廣州510650)


此內(nèi)容為AET網(wǎng)站原創(chuàng),未經(jīng)授權(quán)禁止轉(zhuǎn)載。