Directionality of image plays a very important role in human visual system and it is important prior information of image. In this paper we propose a weighted directional total variation model to reconstruct image fro...Directionality of image plays a very important role in human visual system and it is important prior information of image. In this paper we propose a weighted directional total variation model to reconstruct image from its finite number of noisy compressive samples. A novel self-adaption, texture preservation method is designed to select the weight. Inspired by majorization-minimization scheme, we develop an efficient algorithm to seek the optimal solution of the proposed model by minimizing a sequence of quadratic surrogate penalties. The numerical examples are performed to compare its performance with four state-of-the-art algorithms. Experimental results clearly show that our method has better reconstruction accuracy on texture images than the existing scheme.展开更多
A novel adaptive illumination normalization approach is proposed to eliminate the effects caused by illumination variations for face recognition. The proposed method divides an image into blocks and performs discrete ...A novel adaptive illumination normalization approach is proposed to eliminate the effects caused by illumination variations for face recognition. The proposed method divides an image into blocks and performs discrete cosine transform(DCT) in blocks independently in the logarithm domain. For each block-DCT coefficient except the direct current(DC) component, we take the illumination as main signal and take the reflectance as "noise". A data-driven and adaptive soft-thresholding denoising technique is employed in each block-DCT coefficient except the DC component. Illumination is estimated by applying the inverse DCT in the block-DCT coefficients, and the indirectly obtained reflectance can be used in further recognition task. Experimental results show that the proposed approach outperforms other existing methods. Moreover, the proposed method does not need any prior information, and none of the parameters can be determined by experience.展开更多
基金the National Natural Science Foundation of China(Nos.11401318 and 11671004)the Natural Science Foundation of the Jiangsu Higher Education Institutions of China(No.15KJB110018)the Scientific Research Foundation of NUPT(No.NY214023)
文摘Directionality of image plays a very important role in human visual system and it is important prior information of image. In this paper we propose a weighted directional total variation model to reconstruct image from its finite number of noisy compressive samples. A novel self-adaption, texture preservation method is designed to select the weight. Inspired by majorization-minimization scheme, we develop an efficient algorithm to seek the optimal solution of the proposed model by minimizing a sequence of quadratic surrogate penalties. The numerical examples are performed to compare its performance with four state-of-the-art algorithms. Experimental results clearly show that our method has better reconstruction accuracy on texture images than the existing scheme.
基金the Natural Science Foundation of Jiangsu Province(No.BK20150784)the Fund of Jiangsu Key Laboratory of Image and Video Understanding for Social Safety(Nanjing University of Science and Technology)(No.30920140122007)
文摘A novel adaptive illumination normalization approach is proposed to eliminate the effects caused by illumination variations for face recognition. The proposed method divides an image into blocks and performs discrete cosine transform(DCT) in blocks independently in the logarithm domain. For each block-DCT coefficient except the direct current(DC) component, we take the illumination as main signal and take the reflectance as "noise". A data-driven and adaptive soft-thresholding denoising technique is employed in each block-DCT coefficient except the DC component. Illumination is estimated by applying the inverse DCT in the block-DCT coefficients, and the indirectly obtained reflectance can be used in further recognition task. Experimental results show that the proposed approach outperforms other existing methods. Moreover, the proposed method does not need any prior information, and none of the parameters can be determined by experience.
