-
摘要:
针对像素缺陷影响电润湿电子纸的显示效果,本文提出一种基于Otsu的自动阈值检测方法对缺陷进行检测。Otsu是一种常用的自动阈值方法,在图像直方图为双峰时,该方法能给出令人满意的结果。但是电润湿缺陷图像直方图通常为单峰,容易得到错误的结果。电润湿由于不同颜色的填充油墨使得缺陷与背景对比度不同,导致分割更加困难。本文在目标方差前引入加权系数,权值随着缺陷的累积概率的增大而减小。权值在阈值过峰前保持较大的数值,过峰后权值降低,保证了阈值在单峰情况下始终处于峰的左边。实验结果表明:本文提出的方法能够有效分割电润湿缺陷区域,尤其在对比度较低的电润湿缺陷图像中比Otsu、VE、WOV和熵加权方法的ME值更接近0,有更好的分割效果。
Abstract:Aiming at the effect of pixel defects on the display of electrowetting electronic paper, an automatic threshold detection method based on Otsu is proposed to detect defects. Otsu is a commonly used automatic threshold method that gives satisfactory results when the image histogram is bimodal. However, the electrowetting defect image histogram is usually a single peak, and Otsu method fails. Electrowetting differs from the background contrast due to the filling inks of different colors, making segmentation more difficult. In this paper, the weighting coefficient is introduced before the target variance, and the weight decreases as the cumulative probability of defects increases. The weight keeps a large value before the threshold crosses the peak, and the weight decreases after the peak, ensuring that the threshold is always to the left of the peak in the case of a single peak. The experimental results show that the proposed method can effectively segment the electrowetting defect region, especially in the electrowetting defect image with lower contrast ratio. The method is closer to 0 compared to the ME value of Otsu, VE, WOV and entropy weighting methods. The proposed method has a better segmentation effect.
-
Key words:
- electrowetting display /
- image processing /
- machine vision /
- defect segmentation /
- Otsu method
-
Overview: Electrowetting is an electronic paper display with the advantages of fast response, low cost, and low energy loss. Electrowetting display technology is currently in a high-speed development period. Electrowetting devices may have defects in the production process, due to ink spillage, the uneven spin coating of the hydrophobic insulating layer, and externally introduced impurities. Defects affect the display of electrowetting devices, so the detection of defects is indispensable. Aiming at the effect of pixel defects on the display of electrowetting electronic paper, an automatic threshold detection method based on Otsu is proposed to detect defects. Otsu is a commonly used automatic threshold method that gives satisfactory results when the image histogram is bimodal. However, since the electrowetting defect image histogram is usually unimodal, it is easy to get incorrect results. Moreover, the electrowetting electronic paper is filled with pixels by three primary inks to realize color display, so the contrast between the defect and the background is different, and the difficulty of segmentation is also different. In order to separate the electrowetting defects in different shades of ink, an improved Otsu threshold segmentation algorithm is proposed. The basic principle of the method is to introduce a weighting factor before the target variance and affect the value of the variance between-class by the weighting factor, which affects the final threshold selection. The weight decreases with the increase of the probability of accumulation of defects, and the weight is always at a higher value when the threshold is at the left edge of the peak. Specifically, the weights have different change rules before and after the peak to change the contribution of the defect variance. When the threshold passes through the peak, the contribution of the defect variance is reduced, and the value of the between-class variance is affected by the weight, which can make the threshold on the left side of the peak. The experimental results show that the proposed method can effectively segment the electrowetting defect area. In the electrowetting defect image in the dark ink with low contrast, the ME value of this method is small. Otsu and other automatic threshold methods have ME values above 0.87, and the segmentation results are far from the desired threshold. The proposed method can segment defects in different inks and have better segmentation effects especially when the contrast of defects and background is low.
-
图 5 缺陷图像与灰度方差图。(a)原始图像;(b) Otsu结果;(c)本文方法分割结果;(d)灰度方差曲线;(e)灰度方差曲线与加权后缺陷方差曲线;(f) Otsu类间方差曲线与加权后类间方差曲线
Figure 5. Defect image and histogram. (a) Original image; (b) Otsu method; (c) Proposed method; (d) Gray variance curve; (e) Gray variance curve and weighted defect variance curve; (f) Between-class variance of Otsu and weighted between-class variance curve
图 6 五种自动阈值方法对电润湿图像Ⅰ的分割结果。(a)原始图像;(b) VE方法;(c) Otsu方法;(d) WOV方法;(e) EW方法;(f)本文提出方法;(g)直方图和阈值
Figure 6. Segmentation results of electrowetting image Ⅰ by five automatic threshold methods. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) Proposed method; (g) Histogram and threshold
图 7 五种自动阈值方法对电润湿图像Ⅱ的分割结果。(a)原始图像;(b) VE方法;(c) Otsu方法;(d) WOV方法;(e) EW方法;(f)本文提出方法;(g)直方图和阈值
Figure 7. Segmentation results of electrowetting image Ⅱ by five automatic threshold methods. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) Proposed method; (g) Histogram and threshold
图 8 五种自动阈值方法对电润湿图像Ⅲ的分割结果。(a)原始图像;(b) VE方法;(c) Otsu方法;(d) WOV方法;(e) EW方法;(f)本文提出方法;(g)直方图和阈值
Figure 8. Segmentation results of electrowetting image Ⅲ by five automatic threshold methods. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) Proposed method; (g) Histogram and threshold
图 9 五种自动阈值方法对电润湿图像Ⅳ的分割结果。(a)原始图像;(b) VE方法;(c) Otsu方法;(d) WOV方法;(e) EW方法;(f)本文提出方法;(g)直方图和阈值
Figure 9. Segmentation results of electrowetting image Ⅳ by five automatic threshold methods. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) Proposed method; (g) Histogram and threshold
图 10 五种自动阈值方法对电润湿图像Ⅴ的分割结果。(a)原始图像;(b) VE方法;(c) Otsu方法;(d) WOV方法;(e) EW方法;(f)本文提出方法;(g)直方图和阈值
Figure 10. Segmentation results of electrowetting image Ⅴ by five automatic threshold methods. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) Proposed method; (g) Histogram and threshold
表 1 五种方法分割结果的ME值
Table 1. ME value of processing result by five methods
Electrowetting defect image Otsu VE WOV EW Proposed method Image Ⅰ 0.0294 0.0289 0.0253 0.8393 0.0239 Image Ⅱ 0.6663 0.8307 0.0270 0.9677 0.0035 Image Ⅲ 0.8794 0.8783 0.8830 0.8753 0.0118 Image Ⅳ 0.8363 0.8363 0.8453 0.8343 0.0465 Image Ⅴ 0.8547 0.8547 0.8585 0.8542 0.0144 -
[1] Hayes R A, Feenstra B J. Video-speed electronic paper based on electrowetting[J]. Nature, 2003, 425(6956): 383–385. doi: 10.1038/nature01988
[2] Overton G. ELECTRONIC PAPER DISPLAYS: Kindles and cuttlefish: Biomimetics informs e-paper displays[J]. Laser Focus World, 2012, 48(12): 16. http://d.old.wanfangdata.com.cn/NSTLQK/NSTL_QKJJ0228428854/
[3] Hayes R A, Feenstra B J, Camps I G J, et al. 52.1: a high brightness colour 160 PPI reflective display technology based on electrowetting[J]. SID Symposium Digest of Technical Papers, 2004, 35(1): 1412–1415. doi: 10.1889/1.1825770
[4] Cheng W Y, Lo K L, Chang Y P, et al. 37.1: novel development of large-sized electrowetting display[J]. SID Symposium Digest of Technical Papers, 2008, 39(1): 526–529. doi: 10.1889/1.3069718
[5] Kuo S W, Chang Y P, Cheng W Y, et al. 34.3: novel development of multi-color electrowetting display[J]. SID Symposium Digest of Technical Papers, 2009, 40(1): 483–486. doi: 10.1889/1.3256821
[6] Schultz A, Heikenfeld J, Kang H S, et al. 1000:1 contrast ratio transmissive electrowetting displays[J]. Journal of Display Technology, 2011, 7(11): 583–585. doi: 10.1109/JDT.2011.2160842
[7] Chang R L J, Liu P W, Wu C Y, et al. 54.2: reliable and high performance transparent electrowetting displays[J]. SID Symposium Digest of Technical Papers, 2014, 45(1): 785–788. doi: 10.1002/j.2168-0159.2014.tb00206.x
[8] Zhang X M, Bai P F, Hayes R A, et al. Novel driving methods for manipulating oil motion in electrofluidic display pixels[J]. Journal of Display Technology, 2016, 12(2): 200–205. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=b7cce66d2ef188d3e4d7ea5b51137625
[9] 赵瑞, 田志强, 刘启超, 等.介电润湿液体光学棱镜[J].光学学报, 2014, 34(12): 1223003. http://d.old.wanfangdata.com.cn/Periodical/hbydjs201205003
Zhao R, Tian Z Q, Liu Q C, et al. Electrowetting-based liquid prism[J]. Acta Optica Sinica, 2014, 34(12): 1223003. http://d.old.wanfangdata.com.cn/Periodical/hbydjs201205003
[10] 钱明勇, 林珊玲, 曾素云, 等.电润湿电子纸的实时动态显示驱动系统实现[J].光电工程, 2019, 46(6): 180623. doi: 10.12086/oee.2019.180623
Qian M Y, Lin S L, Zeng S Y, et al. Real-time dynamic driving system implementation of electrowetting display[J]. Opto-Electronic Engineering, 2019, 46(6): 180623. doi: 10.12086/oee.2019.180623
[11] Jin S Q, Ji C, Yan C C, et al. TFT-LCD mura defect detection using DCT and the dual-γ piecewise exponential transform[J]. Precision Engineering, 2018, 54: 371–378. doi: 10.1016/j.precisioneng.2018.07.006
[12] 何俊杰, 肖可, 刘畅, 等.基于区域神经网络的TFT-LCD电路缺陷检测方法[J].计算机与现代化, 2018(7): 33–38. doi: 10.3969/j.issn.1006-2475.2018.07.007
He J J, Xiao K, Liu C, et al. TFT-LCD circuit defects detection based on faster R-CNN[J]. Computer and Modernization, 2018(7): 33–38. doi: 10.3969/j.issn.1006-2475.2018.07.007
[13] Otsu N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on Systems, Man, and Cybernetics, 1979, 9(1): 62–66. doi: 10.1109/TSMC.1979.4310076
[14] Ng H F. Automatic thresholding for defect detection[J]. Pattern Recognition Letters, 2006, 27(14): 1644–1649. doi: 10.1016/j.patrec.2006.03.009
[15] Fan J L, Lei B. A modified valley-emphasis method for automatic thresholding[J]. Pattern Recognition Letters, 2012, 33(6): 703–708. doi: 10.1016/j.patrec.2011.12.009
[16] 张博, 倪开灶, 王林军, 等.基于背景校正和图像分割定量分析光学元件表面疵病的新算法[J].光学学报, 2016, 36(9): 0911004. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=gxxb201609015
Zhang B, Ni K Z, Wang L J, et al. New algorithm of detecting optical surface imperfection based on background correction and image segmentation[J]. Acta Optica Sinica, 2016, 36(9): 0911004. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=gxxb201609015
[17] Yuan X C, Lu W S, Peng Q J. An improved Otsu method using the weighted object variance for defect detection[J]. Applied Surface Science, 2015, 349: 472–484. doi: 10.1016/j.apsusc.2015.05.033
[18] Truong M T N, Kim S. Automatic image thresholding using Otsu's method and entropy weighting scheme for surface defect detection[J]. Soft Computing, 2018, 22(13): 4197–4203. doi: 10.1007/s00500-017-2709-1
[19] Liao P S, Chen T S, Chung P C. A fast algorithm for multilevel thresholding[J]. Journal of Information Science and Engineering, 2001, 17(5): 713–727. http://d.old.wanfangdata.com.cn/OAPaper/oai_arXiv.org_cs%2f0602044
[20] Yasnoff W A, Mui J K, Bacus J W. Error measures for scene segmentation[J]. Pattern Recognition, 1977, 9(4): 217–231. doi: 10.1016/0031-3203(77)90006-1