| 黎双文，梁 耀，刘鑫磊，李丰果.复杂背景下目标树叶的分割与补缺[J].广东农业科学,2018,45(5):120-126 |
|
查看全文
HTML
复杂背景下目标树叶的分割与补缺 |
| Segmentation and replenishment of the target leaves in complex background |
| |
| DOI：10.16768/j.issn.1004-874X.2018.05.020 |
| 中文关键词: 图像分割 复杂背景 超绿算法 模糊C 聚类 凸包填充 |
| 英文关键词: image segmentation complex background extra-green algorithm fuzzy C-cluster algorithm convex hull filling |
| 基金项目:广东省自然科学基金（2016A030313433） |
|
| 摘要点击次数: 1801 |
| 全文下载次数: 684 |
| 中文摘要: |
| 为了进一步提高复杂背景下目标树叶分割算法的运行效率和准确性，在超绿算法的基础上结
合模糊C 聚类算法提高分割的运行效率，利用凸包填充法提高分割的准确性。首先利用超绿算法去除复
杂背景中的非绿色部分，然后利用模糊C 聚类算法去除与目标树叶颜色差异较大的绿色背景，而对于颜
色差异较小的绿色背景，则先利用底帽变换得到目标树叶的边缘信息，再通过腐蚀和比较连通区域的大小
进行去除。采用凸包填充法对目标树叶的边缘缺口进行补缺可以提高分割的准确性，从而降低错分率。超
绿算法结合模糊C 聚类算法可提高分割的运行效率。研究结果表明，与原有算法对比，错分率平均降低了
1.05%，分割效率平均提高了13.82%。 |
| 英文摘要: |
| In order to furtherly improve the operation efficiency and accuracy of the segmentation algorithm of
target leaves in complex background，fuzzy C-cluster algorithm was adopted to improve the operation efficiency of
the segmentation on the basis of the extra-green algorithm，and the convex hull filling method was used to improve
the accuracy of the segmentation. First of all，the non-green part of the complex background was removed by using
the extra-green algorithm. The green background which was largely different from the color of the target leaves was
removed by using fuzzy C-cluster algorithm. While for the green background with smaller color difference，the edge
information of the target leaves was obtained by using bottom- hat transformation，and then the area was removed
by corrosion and the comparison of the connected regions. The edge gap of the target leaves was replenished to
improve the segmentation accuracy by convex hull filling，reducing the rate of error rates. In addition，the operation
efficiency of segmentation could be increased by extra-green algorithm combined with fuzzy C-cluster algorithm. The
experimental results showed that the rate of error rate was by an average of 1.05%，and compared with the original
algorithm，the segmentation efficiency was increased by 13.82% averagely. |
|
查看/发表评论 下载PDF阅读器 |
