| 摘要: |
| 针对国内外苹果分选机在利用计算机图像识别系统进行苹果分级时，需要对每个苹果采集多帧图片，不仅耗时长，而且对梗萼凹陷与苹果真实缺陷易于混淆而影响分级精度的问题，设计了一种苹果分选机输送定位机构。该机构在系统进行图像采集之前，首先实现果梗垂直于水平面的定位，苹果平动输送过程中计算机视觉系统进行苹果图像采集和处理。由于在图像处理时无需再对梗萼凹陷进行识别，并简化了最大直径的确定过程，从而减少了图像处理时间。并可提高识别精度。试验结果表明，该定位机构对每个苹果的定位耗时多在4S之内完成，定位成功率达94％，可满足苹果在线实时检测的需求。 |
| 关键词: 计算机视觉 苹果分选 输送定位机构 |
| DOI：10.11841/j.issn.1007-4333.2005.05.128 |
| 投稿时间：2004-10-07 |
| 基金项目:新西兰ASIA DAF（亚洲发展援助基金）项目“Fruit Quality Assurance System Development”资助项目 |
|
| Study on conveying and orientating section of an apple sorter |
|
|
|
| Abstract: |
| A conveying and orientating section of an apple sorter was developed aiming at the problems of confusion of stem and calyx and grading errors encountered in a common sorter equipped with computer vision system. When this mechanism is working, the apple can be oriented by the stem-calyx in vertical direction. The experiments indicated that the mechanism oriented apples succeed by 94%, took less than 4 s for each apple, and met the requirements of a online detection. After the orientation, apples move with the conveyor while the computer vision system collects and processes images. It takes less time to process images and improves the precision for no distinguish of stem-calyx defects. |
| Key words: computer vision apple sorting orientating and conveying mechanisms |
