Micro-image definition evaluation using multi-scale decomposition and gradient absolute value
-
摘要:
针对显微图像自动对焦和成像系统质量评价问题,结合多尺度分解工具和梯度绝对值算子设计,提出了一种显微图像清晰度评价算法。采用非下采样剪切波分解,对输入的显微图像进行多尺度、多方向变换,得到一幅低频子带图像和若干幅高频子带图像。结合抗噪阈值设置,计算各子带图像的梯度绝对值算子和,利用图像清晰度变化对于低频和高频子带系数影响的差异,将高低频梯度绝对值算子的比值作为最终的显微图像清晰度评价数值。开展了仿真论证实验和实拍论证实验,实验结果表明,所提出的清晰度评价算法具有较好的单调性和抗噪特性,和几种经典的评价算法相比,本文方法得到的评价结果在灵敏度、稳定性和鲁棒性方面表现更为优异,具有很好的实际应用价值。
Abstract:Aimed at the problem of automatic focus and image system quality evaluation in microscopy imaging, a micro-image definition evaluation method is presented by combining multi-scale decomposition tools and absolute gradient operators. The multiscale and multidirectional non-subsampled Shearlet transform is utilized to decompose the input micro image into a low frequency sub-band image and a number of high frequency sub-band images. Combined with the anti-noise threshold setting, the gradient absolute sum values of each sub-band image were calculated. By using the different effects of image sharpness on the low-frequency and high-frequency sub-band coefficients, the ratio of the high-frequency to low-frequency gradient absolute value operator was taken as the final evaluation value of the microscopic image sharpness. The simulation experiment and actual experiments were carried out and the experimental results illustrated that the proposed approach has good monotonicity and anti-noise characteristics. Compared with other classic evaluation algorithms, the presented method obtained superior performance on sensitivity, stability and robustness. It has very good practical application values.
-
Overview: As an important instrument for observing the micro world, optical microscope has been widely used in medical health, biological detection, industrial production and other related fields. The evaluation and determination of micro-image clarity has a direct impact on the accuracy of microscopy autofocus and has become an important index to measure the imaging quality of microscopy system. With the development of multimedia technology and digital image, the requirements for automation of microscopic instrument and equipment have been gradually improved, and more and more attention has been paid to the image process-based image sharpness evaluation algorithm, which is of great significance for realizing rapid and accurate microscopic autofocus and imaging system performance evaluation.
In this paper, a micro-image definition evaluation method is presented by combining multi-scale decomposition tools and absolute gradient operators. The non-subsampled Shearlet transform (NSST) is utilized to decompose the input micro image into a low frequency sub-band image and a number of high frequency sub-band images. NSST is a very effective image multi-scale decomposition tool proposed in recent years. Its mathematical structure is simple and has the characteristics of parabola scale, stronger directional sensitivity and optimal sparse. It can better express the image contour, edge, texture and other detail, which is suitable for image feature extraction and can provide more judgment information for the sharpness evaluation algorithm. Meanwhile, combined with the anti-noise threshold setting, the sum of gradient absolute (SAG) values of each sub-band image was calculated. The SAG operator replaces the square operator in the energy gradient function with absolute value calculation, which reduces the complexity of the calculation and improves the operation efficiency while representing the edge clarity of the image. At last, by using the different effects of image sharpness on the low-frequency and high-frequency sub-band coefficients, the ratio of the high-frequency to low-frequency gradient absolute value operator was taken as the final evaluation value of the microscopic image sharpness. In order to verify the performance of the algorithm, the simulation experiment and actual experiments were carried out. Image sequences with different degrees of blur were simulated and captured and several compared image sharpness evaluation methods were used to give out objective evaluation index values for these image sequences. The experimental results illustrated that the proposed approach has good monotonicity and anti-noise characteristics. Compared with other classic evaluation algorithms, the presented method obtained superior performance on sensitivity, stability and robustness. It has very good practical application values.
-
图 3 不同模糊程度的显微图像仿真结果。(a)清晰图;(b)高斯模糊图σ=1; (c)高斯模糊图σ=2; (d)高斯模糊图σ=3; (e)高斯模糊图σ=4; (f)高斯模糊图σ=5
Figure 3. Simulation results for microscopic images of different image blur.(a) Sharp image; (b) Gaussian blurred image σ=1; (c) Gaussian blurred image σ=2; (d) Gaussian blurred image σ=3; (e) Gaussian blurred image σ=4; (f) Gaussian blurred image σ=5
表 1 几种对比方法对于仿真显微图像序列的评价结果
Table 1. Evaluation results for simulated microscopic image sequence of several compared methods
模糊核标准差 Tenengrad EOG LS GMG AG Proposed σ=0 3.749x103 87.43 9.257 2.111 3.771 1.749x10-1 σ=1 2.484x103 42.48 3.536 1.472 2.585 1.065x10-1 σ=2 1.598x103 25.67 1.753 1.144 2.228 5.082x10-2 σ=3 1.189x103 18.82 1.163 0.9793 2.107 1.823x10-2 σ=4 1.014x103 15.97 0.9316 0.9022 2.059 4.705x10-3 σ=5 9.356x102 14.72 0.8342 0.8661 2.037 1.148x10-3 表 2 几种清晰度评价方法灵敏度参数比较
Table 2. Comparisons of sensitivity parameter for several definition evaluation methods
评价方法 Tenengrad EOG LS GMG AG Proposed 灵敏度数值 0.750 0.832 0.910 0.590 0.460 0.994 表 3 植物细胞有丝分裂图像评价结果
Table 3. Evaluation results of plant cell mitosis image
图像序列 Tenengrad EOG LS GMG AG Proposed 1 122.9 3.717 4.720 0.4354 2.410 2.147x10-3 2 235.4 5.925 5.227 0.5497 2.891 4.465x10-3 3 358.5 8.412 5.706 0.6549 3.386 1.368x10-2 4 658.1 14.56 6.719 0.8618 4.455 2.166x10-2 5 1246 27.40 8.824 1.1820 5.860 4.892x10-2 6 1493 33.30 10.04 1.3031 5.749 5.367x10-2 7 664.9 14.97 7.062 0.8736 4.066 1.530x10-2 8 393.3 9.239 5.968 0.6864 3.369 7.278x10-3 9 255.0 6.388 5.370 0.5708 2.916 5.278x10-3 10 183.1 4.926 5.026 0.5012 2.641 2.070x10-3 11 136.7 4.000 4.806 0.4516 2.443 2.061x10-3 表 4 洋葱表皮细胞图像评价结果
Table 4. Evaluation results of onion scale epidermal cell image
图像序列 Tenengrad EOG LS GMG AG Proposed 1 299.0 7.304 5.494 0.6103 4.862 2.438x10-3 2 477.9 11.00 6.198 0.7489 5.245 5.652x10-3 3 587.6 13.36 6.642 0.8255 5.627 1.137x10-2 4 767.2 17.37 7.347 0.9411 6.169 3.526x10-2 5 919.7 20.99 8.023 1.035 6.551 4.223x10-2 6 732.7 16.89 7.368 0.9279 6.001 3.263x10-2 7 539.7 12.60 6.604 0.8017 5.379 1.764x10-2 8 381.5 9.156 5.936 0.6832 4.816 6.008x10-3 9 300.5 7.470 5.588 0.6172 4.493 2.877x10-3 10 231.7 6.027 5.276 0.5544 4.178 1.896x10-3 表 5 实拍显微图像清晰度评价方法灵敏度参数比较
Table 5. Comparisons of sensitivity parameter for several definition evaluation methods of actual micro-images
评价方法 Tenengrad EOG LS GMG AG Proposed 植物细胞有丝分裂图 0.918 0.889 0.530 0.666 0.589 0.958 洋葱鳞片表皮细胞图 0.748 0.713 0.342 0.464 0.362 0.955 -
[1] Ilhan H A, Dogar M, Ozcan M. Digital holographic microscopy and focusing methods based on image sharpness[J]. Journal of Microscopy, 2014, 255(3): 138-149. doi: 10.1111/jmi.12144
[2] Rudnaya M E, Mattheij R M M, Maubach J M L. Evaluating sharpness functions for automated scanning electron microscopy[J]. Journal of Microscopy, 2010, 240(1): 38-49. doi: 10.1111/jmi.2010.240.issue-1
[3] 郑馨, 艾列富, 刘奎, 等.结合全局和局部灰度变化的显微图像自动聚焦函数[J].激光与光电子学进展, 2017, 54(8): 081801. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201708032
Zheng X, Ai L F, Liu K, et al. Auto-focusing function for microscopic images based on global and local gray-scale variation[J]. Laser & Optoelectronics Progress, 2017, 54(8): 081801. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jgygdzxjz201708032
[4] 赵巨峰, 毛磊, 刘承, 等.视觉注意机制与边缘展宽衡量相结合的显微成像清晰度评价[J].光子学报, 2015, 44(7): 0711002. http://d.old.wanfangdata.com.cn/Periodical/gzxb201507021
Zhao J F, Mao L, Liu C, et al. Microscopy imaging definition criterion using visual attention mechanism and edge spreading evaluation[J]. Acta Photonica Sinica, 2015, 44(7): 0711002. http://d.old.wanfangdata.com.cn/Periodical/gzxb201507021
[5] 周厚奎, 葛品森, 冯海林.基于非下采样Contourlet变换的显微图像清晰度评价算法[J].合肥工业大学学报(自然科学版), 2014, 37(10): 1204-1209. doi: 10.3969/j.issn.1003-5060.2014.10.011
Zhou H K, Ge P S, Feng H L. Micro-image definition evaluation algorithm based on non-subsampled Contourlet transform[J]. Journal of Hefei University of Technology (Natural Science), 2014, 37(10): 1204-1209. doi: 10.3969/j.issn.1003-5060.2014.10.011
[6] 江旻珊, 张楠楠, 张学典, 等.混合搜索法在显微镜自动对焦中的应用[J].光电工程, 2017, 44(7): 685-694. doi: 10.3969/j.issn.1003-501X.2017.07.004
Jiang M S, Zhang N N, Zhang X D, et al. Applications of hybrid search strategy in microscope autofocus[J]. Opto-Electronic Engineering, 2017, 44(7): 685-694. doi: 10.3969/j.issn.1003-501X.2017.07.004
[7] 黄德天.基于图像技术的自动调焦方法研究[D].长春: 中国科学院研究生院(长春光学精密机械与物理研究所), 2013.
Huang D T. Study on auto-focusing method using image technology[D]. Changchun: Graduate School of Chinese Academy of Sciences (Changchun Institute of Optical Precision Machinery and Physics), 2013.
http://ir.ciomp.ac.cn/handle/181722/35793 [8] 王平江, 陈德军, 巫孟良, 等.一种复合的自动对焦方法在影像测量仪中的应用[J].中国机械工程, 2007, 18(21): 2555-2560. doi: 10.3321/j.issn:1004-132x.2007.21.010
Wang P J, Chen D J, Wu M L, et al. Implementation of a multiple automatic focusing algorithm for video measuring system[J]. China Mechanical Engineering, 2007, 18(21): 2555-2560. doi: 10.3321/j.issn:1004-132x.2007.21.010
[9] Jin H Y, Wang Y Y. A fusion method for visible and infrared images based on contrast pyramid with teaching learning based optimization[J]. Infrared Physics & Technology, 2014, 64: 134-142. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=e41fd98a492f4cae547ee02a0c245d01
[10] 郭敬滨, 冯华杰, 王龙, 等.基于梯度能量函数的调焦窗口构建方法[J].红外技术, 2016, 38(3): 197-202. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=hwjs201603004
Guo J B, Feng H J, Wang L, et al. Design of focusing window based on energy function of gradient[J]. Infrared Technology, 2016, 38(3): 197-202. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=hwjs201603004
[11] 姚松, 曹丹华, 吴裕斌.图像信息量的变化对自动对焦评价函数的影响[J].光电工程, 2006, 33(5): 81-84, 90. doi: 10.3969/j.issn.1003-501X.2006.05.018
Yao S, Cao D H, Wu Y B. Performance of autofocus judgements in different quantity of information series images[J]. Opto-Electronic Engineering, 2006, 33(5): 81-84, 90. doi: 10.3969/j.issn.1003-501X.2006.05.018
[12] 张亚涛, 吉书鹏, 王强锋, 等.基于区域对比度的图像清晰度评价算法[J].应用光学, 2012, 33(2): 293-299. http://d.old.wanfangdata.com.cn/Periodical/yygx201202012
Zhang Y T, Ji S P, Wang Q F, et al. Definition evaluation algorithm based on regional contrast[J]. Journal of Applied Optics, 2012, 33(2): 293-299. http://d.old.wanfangdata.com.cn/Periodical/yygx201202012
[13] 金伟正, 冷秋君, 张卓, 等.基于Contourlet变换的多尺度图像质量评价[J].武汉大学学报(理学版), 2015, 61(2): 192-196. http://d.old.wanfangdata.com.cn/Periodical/whdxxb-zr201502016
Jin W Z, Leng Q J, Zhang Z, et al. Contourlet transform based multiscale image quality assessment metric[J]. Journal of Wuhan University (Natural Science Edition), 2015, 61(2): 192-196. http://d.old.wanfangdata.com.cn/Periodical/whdxxb-zr201502016
[14] Wu W, Qiu Z M, Zhao M, et al. Visible and infrared image fusion using NSST and deep Boltzmann machine[J]. Optik, 2018, 157: 334-342. doi: 10.1016/j.ijleo.2017.11.087
[15] 王烨茹, 冯华君, 徐之海, 等.一种覆盖范围可调的变频梯度自动对焦评价函数[J].红外与激光工程, 2016, 45(10): 1028001. http://d.old.wanfangdata.com.cn/Periodical/hwyjggc201610040
Wang Y R, Feng H J, Xu Z H, et al. An adjustable coverage range autofocus evaluation function using gradient operator with variable frequency[J]. Infrared and Laser Engineering, 2016, 45(10): 1028001. http://d.old.wanfangdata.com.cn/Periodical/hwyjggc201610040