基于深度感知的水下对空图像增强算法研究
|
摘要
由于水体对光线的吸收和散射,造成水下对空所成图像噪声较大,对比度较低,降低了图像的观察效果。在前期研究工作中采用的去噪和对比度拉伸算法没有利用图像的深度信息,使增强后图像边缘细节不够突出。将深度信息引入水下对空图像去噪及对比度拉伸过程中,采用基于深度约束的均值去噪和基于深度感知的对比度拉伸算法处理水下对空图像,有效提升了水下对空图像的深度层次感及对象边缘的质量,为后期水下对空成像的目标探测和识别奠定了理论与技术基础。
Due to water body absorption and scattering of light,water to air imaging noise is to some extend high and the contrast of scene picture is always low. In this case,the viewing effect of the scene is greatly affected. Image depth information was not taken into account for image denoising and contrast stretching algorithms in the former research work,and this makes the enhanced image edge details not to be sufficient. In this paper,the depth information is introduced into image denoising and contrast stretching,and depth-constrained based mean denosing and depth-aware based contrast stretching algorithm are used to process water to air image which effectively improves the depth of water to air imaging and the quality of the edge of the object. It forms theory and technology foundation for target detection and recognition of water to air imaging.
Due to water body absorption and scattering of light,water to air imaging noise is to some extend high and the contrast of scene picture is always low. In this case,the viewing effect of the scene is greatly affected. Image depth information was not taken into account for image denoising and contrast stretching algorithms in the former research work,and this makes the enhanced image edge details not to be sufficient. In this paper,the depth information is introduced into image denoising and contrast stretching,and depth-constrained based mean denosing and depth-aware based contrast stretching algorithm are used to process water to air image which effectively improves the depth of water to air imaging and the quality of the edge of the object. It forms theory and technology foundation for target detection and recognition of water to air imaging.
引文
[1]叶春明.水下对空成像中的波面估计[J].光学与光电技术,2011,9(2):33-36.YE Chun-ming. Wave estimation on through-surface underwater imaging[J]. Optics&Optoelectronic Technology,2011,9(2):33-36.
[2]徐枫,刘爱东,陈宏利.一种水下对空成像图像增强算法[J].光学与光电技术,2007,5(3):59-62.XU Feng,LIU Ai-dong,CHEN Hong-li. Enhanced algorithm of underwater imaging for air targets[J]. Optics&Optoeiectronic Technology,2007,5(3):59-62.
[3] Lee S. An efficient content-based image enhancement in the compressed domain using Retinex theory[J]. IEEE Transactions on Circuits and Systems for Video Technology,2007,17(2):199-213.
[4] Du S,Ward R-K. Adaptive region-based image enhancement method for robust face recognition under variable illumination conditions[J]. IEEE Transactions on Circuits and Systems for Video Technology,2010,99:1-12.
[5] Boudraa A O,Diop E H S. Image contrast enhancement based on 2D teager-kaiser operator[C]. Proceedings of the IEEE International Conference on Image Processing(ICIP),2008:3180-3183.
[6] Xu Q,Jiang H,Scopigno R,et al. A new approach for very dark video denoising and enhancement[C]. Image Processing(ICIP),2010 17th IEEE International Conference,2010:1185-1188.
[7] HE K M,SUN J,TANG X O. Single image haze removal using dark channel prior[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D C:IEEE Computer Society,2009:1956-1963.
[8] Le Roux N,Bengio Y. Representational power of restricted Boltzmann machines and deep belief networks[J]. Neural N Computation,2008,20(6):1631-1649.
[9] HintonG E,Osindero S. A fast learning algorithmfor deep belief nets[J]. Neural Computation,2006,18(7):1527-1554.
[10] Hinton G,Deng L,Yu D,et al. Deep neural networks for acoustic modeling in speech recognition:The shared views of four research groups[J]. Signal Processing Magazine,IEEE,2012,29(6):82-97.
[2]徐枫,刘爱东,陈宏利.一种水下对空成像图像增强算法[J].光学与光电技术,2007,5(3):59-62.XU Feng,LIU Ai-dong,CHEN Hong-li. Enhanced algorithm of underwater imaging for air targets[J]. Optics&Optoeiectronic Technology,2007,5(3):59-62.
[3] Lee S. An efficient content-based image enhancement in the compressed domain using Retinex theory[J]. IEEE Transactions on Circuits and Systems for Video Technology,2007,17(2):199-213.
[4] Du S,Ward R-K. Adaptive region-based image enhancement method for robust face recognition under variable illumination conditions[J]. IEEE Transactions on Circuits and Systems for Video Technology,2010,99:1-12.
[5] Boudraa A O,Diop E H S. Image contrast enhancement based on 2D teager-kaiser operator[C]. Proceedings of the IEEE International Conference on Image Processing(ICIP),2008:3180-3183.
[6] Xu Q,Jiang H,Scopigno R,et al. A new approach for very dark video denoising and enhancement[C]. Image Processing(ICIP),2010 17th IEEE International Conference,2010:1185-1188.
[7] HE K M,SUN J,TANG X O. Single image haze removal using dark channel prior[C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D C:IEEE Computer Society,2009:1956-1963.
[8] Le Roux N,Bengio Y. Representational power of restricted Boltzmann machines and deep belief networks[J]. Neural N Computation,2008,20(6):1631-1649.
[9] HintonG E,Osindero S. A fast learning algorithmfor deep belief nets[J]. Neural Computation,2006,18(7):1527-1554.
[10] Hinton G,Deng L,Yu D,et al. Deep neural networks for acoustic modeling in speech recognition:The shared views of four research groups[J]. Signal Processing Magazine,IEEE,2012,29(6):82-97.