-
图1为水下偏振距离选通成像物理模型示意图。如图1(a)所示,典型的水下偏振距离选通成像系统主要由脉冲激光器、门控成像器件、起偏器、光学发射镜头、检偏器、光学接收镜头、时序控制器(TCU)等组成。工作过程中,TCU控制脉冲激光器和门控成像器件ICMOS按照设计的选通时序工作。其中,脉冲激光器发出激光脉冲,通过起偏器起偏后产生线偏振光,并由光学发射镜头整形后产生所需的偏振脉冲光,偏振脉冲光经目标反射后产生含有目标偏振信息的回波信号;在设定延时下,ICMOS开启选通快门,接收由光学接收镜头采集并由检偏器过滤的目标信号,利用选通空间滤波抑制选通切片与成像系统间传输链路上水体后向散射噪声,并利用偏振滤波进一步抑制选通切片内水体后向散射噪声,实现水下光学图像信噪比和对比度的提升。图1(b)对比示意了成像过程中传统水下光学成像、选通成像和偏振选通成像接收目标信号、水体噪声和背景噪声的差异。传统水下光学成像包含了整个链路上全部的噪声和目标信号,具体包括目标与成像系统间的主要后向散射噪声(main backscatter)、目标附近的残留噪声(sub-backscatter)、目标信号、水体背景噪声;相比传统水下光学成像,选通成像则有效过滤了目标与成像系统间的主要后向散射噪声和水体背景噪声,但是选通图像中还存在选通切片内的残留后向散射噪声;相比选通成像,将光学偏振成像引入选通成像后,偏振选通成像则可进一步抑制选通图像中的残留噪声。需说明的是,成像系统自身噪声在各成像技术中均存在,因此在图1(b)中均未单独示意。偏振选通成像虽然可以进一步抑制选通图像中的残留噪声,但由于起偏器和检偏器对光能量存在衰减,因此在实际应用中,不同水质下选通成像和偏振选通成像需对比研究,以明确适用条件。
王新伟[10]和葛卫龙[11]等对选通成像目标光能量进行了计算,相机接收到的距离r处的目标能量:
$$ E_{r}= \frac{{\rho f}_{L}}{{f}_{C}}{\eta }_{r}{\eta }_{t} \left(\frac{D}{2r}\right)^2 {{\rm{e}}}^{-2cr}\cdot{\int }_{0}^{\mathrm{\infty }}\frac{G\left(t\right)Q\left(t\right)}{{t}_{p}}{\rm{d}}t $$ (1) 式中:fL为脉冲激光重复频率;fC为相机帧频;ρ为水下目标(视为朗伯体)的反射率;G(t)为激光函数;D为接收光学系统直径;ηr和ηt为发射和接收光学系统的效率;Q(t)为门函数;c为水体衰减系数;tp为脉冲宽度。
在对目标信号光进行计算后,需要对水体后向散射光能量进行计算。王新伟[10]和李丽等[12]计算了选通成像过程中的后向散射光能量,综合上述研究成果,文中给出后向散射光能量为:
$$ E_{b}= \frac{{f}_{L}}{{f}_{C}}\left(\frac{D}{2}\right)^2 \sigma \eta_{r}\eta _{t} {\int }_{0}^{\infty }{\int }_{\tfrac{{V}_{w}\left({\tau }_{d}-{t}_{p}\right)}{2}}^{\tfrac{{V}_{w}\left({\tau }_{d}+{t}_{g}\right)}{2}}\frac{{G\left(t\right)Q\left(t\right){\rm{e}}}^{-2cr}}{{\tau }_{p}r^2}{\rm{d}}t{\rm{d}}r $$ (2) 式中:σ为水体后向散射系数;Vw为水中光速;τd为开门时刻(延时);tg为选通门宽。G(t)和Q(t)取值如下:
$$ G(t)= \left\{\begin{array}{c}G,\;{\tau }_{d} \leqslant t \leqslant {{\tau }_{d}+t}_{g}\\ 0,\;{\rm{else}}\;{{t}}\end{array}\right. $$ (3) $$ Q(t)= \left\{\begin{array}{c}Q,\;{\tau }_{d}\leqslant t \leqslant {{\tau }_{d}+t}_{g}\\ 0,\;{\rm{else}}\;{{t}}\end{array}\right. $$ (4) 式中:G为相机增益;Q为激光单脉冲能量。
偏振选通成像无法过滤切片内的前向散射光,在高浑浊水体中,前向散射光对成像质量影响严重,因此,文中在建立偏振选通成像物理模型时计算前向散射光能量[13]:
$$ E_{f}= \dfrac{{f}_{L}}{{f}_{c}}{\eta }_{r}{\eta }_{t}{\int }_{{t}_{g}}^{{t}_{g}+{\tau }_{d}}{\int }_{\tfrac{{V}_{w}\left({\tau }_{d}-{t}_{p}\right)}{2}}^{\tfrac{{V}_{w}\left({\tau }_{d}+{t}_{g}\right)}{2}}\frac{\pi {D}^{4}G\left(t\right)Q\left(t\right)\beta \left(\theta \right)\rho {{\rm{e}}}^{-2cr}}{16{\tau }_{p}r^2r{'}^2} {\rm{d}}t {\rm{d}}r' $$ (5) 式中:r′为前向散射体积元距相机距离。体积散射函数采用Henyey-Greenstein 函数:
$$ \beta(\theta)=\frac{b}{4 \pi}\left(1-g^2\right)\left(1+g^2-2 g \cos \theta\right)^{-{2}/{3}} $$ (6) 式中:b为总散射系数;g为非对称因子。取b=0.45c,g=0.8,前向截获因子取1[14-15]。
距离选通成像信噪比如下:
$$ S NR_{RGI}= \frac{{E}_{r}}{{{E}_{d}+E}_{b}+{E}_{f}} $$ (7) 式中:Ed为系统噪声功率。
当水体散射系数低于0.5 mm−1时,532 nm线偏振光入射水体时产生的后向散射光的线偏振度随散射系数的增加呈e指数衰减,圆偏振度呈e指数上升[16-17]。假设线偏振度为DOPl,圆偏振度为DOPc,已知检偏器可以过滤100%的线偏振态后向散射光和50%的圆偏振态后向散射光。单个偏振器的光衰减为50% [18],定义偏振选通信噪比如下:
$$ S N R_{PRGI}= \frac{0.25{E}_{r}}{0.25[\left(1-0.5{DOP}_{c}-{DOP}_{l}\right){E}_{b}+{E}_{f}]+{E}_{d}} $$ (8) 化简后可得:
$$ S N R_{PRGI}= \frac{{E}_{r}}{\left(1-0.5{DOP}_{c}-{DOP}_{l}\right){E}_{b}+{E}_{f}+4{E}_{d}} $$ (9) 由于较难获得准确的水体后向散射光偏振度数据,假设当水体散射系数远低于0.5 mm−1时,后向散射光的线偏振度近似为1,圆偏振度为0,即偏振器能完全过滤后向散射光[16-17]。此外,因模拟深海全黑光照环境,可忽略环境光影响。
文中基于相机接收的目标光和水体散射光能量建立信噪比模型,通过反射率ρ来区分不同目标,与目标频率特性无关。
Influence of optical polarization on underwater range-gated imaging for target recognition distance under different water quality conditions
-
摘要: 水下光学成像技术对于海底资源勘测、海洋生态监测、水下搜索救援、水下考古等应用具有重要意义。相比传统水下摄像机,距离选通成像技术可以过滤选通切片外的后向散射噪声和环境背景噪声,实现高质量水下成像,但是在浑浊水体中仍然会受切片内后向散射噪声影响,导致成像距离缩短。对此,开展了光学偏振与距离选通成像结合的水下偏振选通成像技术研究,利用后向散射光良好的保偏性去除选通切片范围内的后向散射噪声,提升目标识别距离。通过理论仿真和实验研究,对比分析了不同水质下距离选通成像和偏振选通成像目标识别距离的差异。发现存在临界衰减系数c0:当水体衰减系数小于等于c0时,光学偏振对于提升距离选通成像工作距离无效果;当水体衰减系数大于c0时,偏振可提升距离选通成像工作距离。实验中还发现,目标反射率会影响临界衰减系数。该研究有利于不同水质下距离选通成像的优化应用。Abstract:
Objective Underwater optical imaging technology is of great significance for applications such as seabed resource exploration, marine ecological monitoring, underwater search and rescue, and underwater archaeology. Compared to traditional underwater cameras, underwater range-gated imaging (RGI) technology can filter out backscattered noise and environmental background noise outside the gated slice, achieving high-quality underwater imaging. However, in turbid water bodies, it is still affected by backscattered noise inside the slice, resulting in a shorter imaging distance. Methods In view of the problem of short RGI distance in highly turbid water bodies, underwater polarization gating imaging technology combining optical polarization and RGI was studied. By utilizing the good polarization preservation of backscattered light, the backscattered noise within the gating slice range was removed, and the target recognition distance was improved (Fig.1). Firstly, a physical model for polarized-range-gated imaging (PRGI) is established, a formula for calculating the signal-to-noise ratio of PRGI is derived, a normalized simulation curve for signal-to-noise ratio is drawn. Subsequently, RGI and PRGI are performed on underwater targets such as fishing nets and corals, and signal-to-noise ratio normalization experimental curves are drawn. The simulation curves and experimental curves are compared and analyzed. Results and Discussions When the water attenuation coefficient is 0.21 m−1, the PRGI recognition distance is about 15 m, and the RGI recognition distance is about 17 m (Fig.2). The reason why the recognition distance of PRGI is smaller than RGI is that under the low water attenuation coefficient, the backscattering of the water body is small, and the absorption effect of the water body plays a major role in limiting the recognition distance. When the water attenuation coefficient is 0.42 m−1, the recognition distance of PRGI is about 8 m, and RGI recognition distance is about 9 m (Fig.3). The gap between the two has narrowed. The reason is that as water attenuation coefficient increases, the backscattering of water increases, and the scattering effect of water will reduce the target recognition distance. When the water attenuation coefficient is 0.63 m−1, the recognition distance of PRGI is between 5.5 m and 6 m, and the recognition distance of RGI is between 5 m and 5.5 m (Fig.4). When the water attenuation coefficient is relatively high, the backscattering effect of water is severe, PRGI can improve the recognition distance compared with RGI. The experimental results of fishing net imaging under the water attenuation coefficient of 0.21 m−1 show that the signal-to-noise ratio of PRGI at 16 m is lower than RGI (Fig.7). The experimental results of fishing net imaging under the water attenuation coefficient of 0.42 m−1 show that the signal-to-noise ratio of PRGI at 9 m is lower than RGI (Fig.8). The experimental results of fishing net imaging under the water attenuation coefficient of 0.63 m−1 show that the signal-to-noise ratio of PRGI at 5.5 m is better than RGI (Fig.9). The experimental results of coral imaging under the water attenuation coefficient of 0.21 m−1 show that PRGI have severe device noise at 19 m, and the signal-to-noise ratio of the image is worse than RGI (Fig.10). The experimental results of coral imaging under the water attenuation coefficient of 0.54 m−1 show that PRGI at 9.5 m is slightly worse than RGI (Fig.11). The coral experiment results under the water attenuation coefficient of 0.89 m−1 show that the signal-to-noise ratio of PRGI at 5 m is better than RGI (Fig.12). Conclusions According to the comparison experiment between fishing nets and coral, there should be a critical attenuation coefficient c01 between 0.42 m−1 and 0.63 m−1. When the water attenuation coefficient is higher than c01, the maximum recognition distance of PRGI of fishing nets is greater than RGI; There is a critical attenuation coefficient c02 between 0.54 m−1 and 0.89 m−1. When the water attenuation coefficient is higher than c02, the maximum recognition distance of PRGI of coral is greater than RGI. Based on the comprehensive simulation and experimental results, the following conclusions can be drawn. 1) There exist a critical attenuation coefficient c0, which determines the applicable water quality for RGI and PRGI. 2) The critical attenuation coefficient c0 is related to the target reflectivity. -
-
[1] Deng X, Wang H, Liu X, et al. State of the art of the underwater image processing methods [C]//2017 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC). IEEE, 2018: 8242429. [2] 曹峰梅, 金伟其, 黄有为等. 水下光电成像技术与装备研究进展(上)—— 水下激光距离选通技术[J]. 红外与激光工程, 2011, 33. (2): 1001-8891. Cao Fengmei, Jin Weiqi, Huang Youwei, et al. Review of underwater opto-electrical imaging technology and equipment (I) —— Underwater laser range-gated imaging technology [J]. Infrared and Laser Engineering, 2011, 33(2): 1001-8891. (in Chinese) [3] Philip Church, Weilin Hou, Georges Fournier, et al. Overview of a hybrid underwater camera system [C]//Proceedings of SPIE, 2014, 9111: 911100. [4] 王新伟, 孙亮, 王敏敏, 杨于清, 周燕. 水下二维及三维距离选通成像去噪技术研究[J]. 红外与激光工程, 2020, 49(2): 0203002. Wang Xinwei, Sun Liang, Wang Minmin, et al. Deblurring methods for underwater 2D and 3D range gatedimaging [J]. Infrared and Laser Engineering, 2020, 49(2): 003002. (in Chinese) [5] Wang M, Wang X, Zhang Y, et al. Range-intensity-profile prior dehazing method for underwater range-gated imaging [J]. Optics Express, 2021, 29(5): 7630-7640. doi: 10.1364/OE.417131 [6] Hu Haofeng, Li Xiaobo, Liu Tiegen. Recent advances in underwater image restoration technique based on polarimetric imaging [J]. Infrared and Laser Engineering, 2019, 48(6): 0603006. (in Chinese) [7] Treibitz T, Schechner Y Y. Active polarization descattering [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(3): 385-399. doi: 10.1109/TPAMI.2008.85 [8] Cheng Q, Wang Y, Li D. Analysis of polarization characteristics of backscattered light and reflected light of underwater active light source [J]. Optical Engineering, 2021, 60: 126105. [9] 秦琳, 陈名松, 阙斐一. 基于距离选通的水下偏振光成像系统的研究[J]. 电子设计工程, 2011, 19(07): 184-186. DOI: 10.14022/j.cnki.dzsjgc.2011.07.001. Qin Lin, Chen Mingsong, Que Feiyi. Investigation of underwater polarization imagine system based on range-gated techniques [J]. Electronic Design Engineering, 2011, 19(7): 184-186. (in Chinese) doi: 10.14022/j.cnki.dzsjgc.2011.07.001 [10] Wang Xinwei, Sun Liang, Liu Yuliang. Impact of echo broadening effect on active range-gated imaging [J]. Chinese Optics Letters, 2012, 10(10): 101101. doi: 10.3788/COL201210.101101 [11] Ge Weilong, Hua Lianghong, Zhang Xiaohui. Signal to noise research in range-gated underwater laser imaging system [J]. Infrared and Laser Engineering, 2013, 42(8): 2022-2026. (in Chinese) [12] 李丽, 高稚允, 王霞等. 水下距离选通成像系统后向散射光的计算[J]. 北京理工大学学报, 2003, 23(4): 4882491. doi: 10.15918/j.tbit1001-0645.2003.04.022 Li Li, Gao Zhiyun, Wang Xia, et al. Optical backscatter calculation for an underwater range-gated imaging system [J]. Transactions of Beijing Institute of Technology, 2003, 23(4): 4882491. (in Chinese) doi: 10.15918/j.tbit1001-0645.2003.04.022 [13] Maffione R A, Spinrad R W, Dana D R, et al. Instrument for underwater measurement of optical backscatter [C]//Proceedings of SPIE, 1991, 1537: 173-184. [14] Liu Quanhua, Weng Fuzhong. Combined Henyey-Greenstein and Rayleigh phase function [J]. Applied Optics, 2006, 45(28): 7475-7479. doi: 10.1364/AO.45.007475 [15] Cornette W M, Shanks J G. Physically reasonable analytic expression for the single-scattering phase function [J]. Applied Optics, 1992, 31(16): 3152-3160. doi: 10.1364/AO.31.003152 [16] Morgan S P, Ridgway M E. Polarization properties of light backscattered from a two layer scattering medium [J]. Optics Express, 2000, 7(12): 395-402. doi: 10.1364/OE.7.000395 [17] 曹先平, 孙萍, 不同波长的散射介质后向散射光偏振度特性[J]. 光子学报 , 2012 , 41(5): 0506-086. Cao Xianping, Sun Ping. Characteristics of degree of polarization of backscattering light in scattering medium at different wavelengths [J]. Acta Photonica Sinica, 2012, 41(5): 608-613. (in Chinese) [18] 石顺祥, 王学恩, 马琳. 物理光学与应用光学 [M]. 西安: 西安电子科技大学出版社, 2014. [19] 白廷柱, 金伟其. 光电成像原理与技术 [M]. 北京: 北京理工大学出版社, 2006.