| 严成宸,金施群,闫真真,胡海兵.结合加权模板差图与双边滤波的TFT-LCD检测算法[J].电子测量与仪器学报,2017,31(9):1434-1440 |
| 结合加权模板差图与双边滤波的TFT-LCD检测算法 |
| TFT LCD detection algorithm combining weighted template difference image and bilateral filtering |
| |
| DOI：10.13382/j.jemi.2017.09.013 |
| 中文关键词: TFT LCD 缺陷检测 差图法 双边滤波 最大熵分割 |
| 英文关键词:TFT LCD defect detection difference image method bilateral filtering maximum entropy segmentation |
| 基金项目:科技部重大仪器专项（2013YQ220749）资助 |
|
|
| Author | Institution |
| Yan Chengchen | Instrument Science and Optoelectric Engineering College, Hefei University of Technology, Hefei 230009, China |
| Jin Shiqun | 2. Academy of Photoelectric Technology, Hefei University of Technology, Hefei 230009, China;
3. Key Laboratory of Special Display Technology of the Ministry of Education, Hefei 230009, China;
4. National Key Laboratory of Advanced Display Technology, Hefei 230009, China |
| Yan Zhenzhen | Instrument Science and Optoelectric Engineering College, Hefei University of Technology, Hefei 230009, China |
| Hu Haibing | 2. Academy of Photoelectric Technology, Hefei University of Technology, Hefei 230009, China;
3. Key Laboratory of Special Display Technology of the Ministry of Education, Hefei 230009, China;
4. National Key Laboratory of Advanced Display Technology, Hefei 230009, China |
|
| 摘要点击次数: 2203 |
| 全文下载次数: 6730 |
| 中文摘要: |
| 为了检测薄膜晶体管液晶显示器（TFT LCD）面板上常见的点缺陷和线缺陷，提出了一种加权模板差图法与双边滤波相结合的缺陷检测算法。选取待检图左上角l×k大小的理想纹理模板，将原图分割为若干相同大小的单元，与加权的模板相减后再合成以去除大部分纹理背景；采用双边滤波抑制残余纹理背景；采用最大熵法分割缺陷；最后提取图像中的缺陷参数。通过对构造的理论缺陷图试验，验证了算法能检测出TFT LCD面板图像上的缺陷，再对针孔、划痕、颗粒等样本缺陷进行试验并成功检出。算法结合了差图法去背景性能优异和双边滤波去噪效果明显的优点，具有良好的准确性和适用性。 |
| 英文摘要: |
| In order to detect common point and line defects on TFT LCD panels, a defect detection algorithm combining weighted template difference image method and bilateral filtering is proposed. The image is divided into several units with the same size, in which an ideal texture template with the size of l×k in the upper left corner is selected. These units are synthesized after subtracting the weighted template in order to remove most of the texture background. The suppression of residual texture background is achieved by bilateral filtering. The maximum entropy threshold segmentation method is used to realize the defect segmentation. Finally, the defect parameters in the image are extracted. It is proved that the algorithm can detect the defects on the TFT LCD panel image through testing the constructed theoretical defects image, and it is verified that the defects of pinhole, scratch and particles are successfully detected. The proposed algorithm combines the advantages of the well background removing performance of the difference image method and the effective denoising of bilateral filtering, which has good accuracy and applicability. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
