组合机床与自动化加工技术
組閤機床與自動化加工技術
조합궤상여자동화가공기술
MODULAR MACHINE TOOL & AUTOMATIC MANUFACTURING TECHNIQUE
2015年
1期
67-70
,共4页
图像目标定位%标准件%三目视觉%轮廓匹配%特征判断%OpenCV
圖像目標定位%標準件%三目視覺%輪廓匹配%特徵判斷%OpenCV
도상목표정위%표준건%삼목시각%륜곽필배%특정판단%OpenCV
image orientation%standard parts%trinocular vision%contour matching%industrial camera%OpenCV
在制造业自动化生产过程中,需要对标准件上某些目标进行定位,从而以此为依据对生产件进行校准。由于标准件为金属物品,且表面粗糙,打光成像后,目标背景复杂,而当前的图像目标定位算法不稳定。对此,文章提出了一个基于OpenCV与三目视觉的标准件定位机制。首先基于三个Basler工业相机实现图像采集;然后基于形态学处理与阈值分割处理得到目标的大致区域,再通过轮廓匹配得到目标的精确坐标,轮廓特征有周长、长宽比、长宽差。最后引入特征判断机制,实现不良检测。最后测试了该机制性能,结果表明:与普通的图像目标定位算法相比,在图像目标特征不明显,且背景复杂时,该机制具有更好的定位与检测效果,准确定位出图像目标的轮廓。
在製造業自動化生產過程中,需要對標準件上某些目標進行定位,從而以此為依據對生產件進行校準。由于標準件為金屬物品,且錶麵粗糙,打光成像後,目標揹景複雜,而噹前的圖像目標定位算法不穩定。對此,文章提齣瞭一箇基于OpenCV與三目視覺的標準件定位機製。首先基于三箇Basler工業相機實現圖像採集;然後基于形態學處理與閾值分割處理得到目標的大緻區域,再通過輪廓匹配得到目標的精確坐標,輪廓特徵有週長、長寬比、長寬差。最後引入特徵判斷機製,實現不良檢測。最後測試瞭該機製性能,結果錶明:與普通的圖像目標定位算法相比,在圖像目標特徵不明顯,且揹景複雜時,該機製具有更好的定位與檢測效果,準確定位齣圖像目標的輪廓。
재제조업자동화생산과정중,수요대표준건상모사목표진행정위,종이이차위의거대생산건진행교준。유우표준건위금속물품,차표면조조,타광성상후,목표배경복잡,이당전적도상목표정위산법불은정。대차,문장제출료일개기우OpenCV여삼목시각적표준건정위궤제。수선기우삼개Basler공업상궤실현도상채집;연후기우형태학처리여역치분할처리득도목표적대치구역,재통과륜곽필배득도목표적정학좌표,륜곽특정유주장、장관비、장관차。최후인입특정판단궤제,실현불량검측。최후측시료해궤제성능,결과표명:여보통적도상목표정위산법상비,재도상목표특정불명현,차배경복잡시,해궤제구유경호적정위여검측효과,준학정위출도상목표적륜곽。
In the process of manufacture automation production, some target positioning on the standard parts was needed, and on this basis to produce a calibration. Because of the standard parts were made of metal objects, and the surface is rough, imaging after polishing, target with complex background. And the current image target localization algorithm is not stable, when the target is very small, which make charac-teristics not obvious, make poor quality of positioning. To solve this, this paper proposes a target-location of standard parts based on opencv and trinocular vision. First of all, based on three industrial camera to real-ize Image acquisition;Then based on threshold segmentation and morphological processing target area is ac-quired, target precise coordinates is obtained by contour matching again, contour features of area, perime-ter, aspect ratio, width is poor. Characteristics determine mechanism is introduced, for detecting adverse. Finally tested in this paper, the mechanism of performance, the results show that, compared with ordinary image target localization algorithm in the image where the target is very small, which make characteristics not obvious, mechanism in this paper has better positioning effect, pinpoint the outline of the image target.