ABSTRACT
Aiming at the problem of the chinese rubbing image segmentation under a denoising algorithm based on deep convolutional neural network is proposed. Document enhancement and binarization is the main pre-processing step in document analysis process. At first, a feed-forward denoising convolutional neural networks as a pre-processing methods for document image has been used for denoise images of additive white Gaussian noise(AWGN). The residual learning mechanism is used to learn the mapping from the noisy image to the residual image between the noisy image and the clean image in the neural network training process. A median filtering has been employed for denoising 'salt and pepper' noise. Given the learned denoising and enhanced image, we compute the adaptive threshold image using local adaptive threshold algorithm and then applies it to produce a binary output image. Experimental results show that combined those algorithms is robust in dealing with non-uniform illuminated, low contrast historic document images in terms of both accuracy and efficiency.
- Huang Z K, Wang F, Xi J M, et al. Binarization Chinese Rubbing Images Using Gaussian Mixture Model{C}//International Conference on Intelligent Computing. Springer, Cham, 2015: 477--482.Google Scholar
- Manmatha R, Rothfeder J L. A scale space approach for automatically segmenting words from historical handwritten documents{J}. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2005 (8): 1212--1225. Google ScholarDigital Library
- Wan J, Wang D, Hoi S C H, et al. Deep learning for content-based image retrieval: A comprehensive study{C}//Proceedings of the 22nd ACM international conference on Multimedia. ACM, 2014: 157--166. Google ScholarDigital Library
- Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks{C}//Advances in neural information processing systems. 2012: 1097--1105. Google ScholarDigital Library
- Jain V, Seung S. Natural image denoising with convolutional networks{C}//Advances in neural information processing systems. 2009: 769--776. Google ScholarDigital Library
- Burger H C, Schuler C J, Harmeling S. Image denoising: Can plain neural networks compete with BM3D?{C}// Computer Vision and Pattern Recognition. IEEE, 2012:2392--2399. Google ScholarDigital Library
- Masci J, Meier U, Cireşan D, et al. Stacked convolutional auto-encoders for hierarchical feature extraction{C}//International Conference on Artificial Neural Networks. Springer, Berlin, Heidelberg, 2011: 52--59. Google ScholarDigital Library
- Zhang K, Chen Y, Chen Y, et al. Beyond a Gaussian Denoiser: Residual Learning of Deep CNN forImageDenoising{J}. IEEE Transactions on Image Processing APublication of theIEEE Signal Processing Society, 2017, 26(7):3142--3155. Google ScholarDigital Library
- Otsu N. A threshold selection method from gray-level histograms{J}. IEEE transactions on systems, man, and cybernetics, 1979, 9(1): 62--66.Google ScholarCross Ref
- W. Niblack, An introduction to digital image processing, Vol. 34, Prentice-Hall Englewood Cliffs, 1986. Google ScholarDigital Library
- Buades A, Coll B, Morel J M. A review of image denoising algorithms, with a new one{J}. Multiscale Modeling & Simulation, 2005, 4(2): 490--530.Google ScholarCross Ref
- Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation{C}//International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015: 234--241.Google Scholar
- Zhang K, Zuo W, Chen Y, et al. Beyond a gaussiandenoiser: Residual learning of deep cnn for image denoising{J}. IEEE Transactions on Image Processing, 2017, 26(7): 3142--3155. Google ScholarDigital Library
- Sauvola J, Pietikäinen M. Adaptive document image binarization{J}. Pattern recognition, 2000, 33(2): 225--236.Google Scholar
- Bradley D, Roth G. Adaptive thresholding using the integral image{J}. Journal of graphics tools, 2007, 12(2): 13--21.Google Scholar
- Huang Z K, Li Z H, Huang H, et al. Comparison of different image denoising algorithms for Chinese calligraphy images{J}. Neurocomputing, 2016, 188: 102--112.Google ScholarCross Ref
- Burger H C, Schuler C, Harmeling S. Learning how to combine internal and external denoising methods{C}//German Conference on Pattern Recognition. Springer, Berlin, Heidelberg, 2013: 121--130.Google Scholar
Index Terms
- Chinese Rubbing Image Binarization based on Deep Learning for Image Denoising
Recommendations
A GAN-based Denoising Method for Chinese Stele and Rubbing Calligraphic Image
AbstractChinese calligraphic images have important artistic and historical values. However, subjected to hundreds of years of natural weathering, corrosion and man-made destruction, Chinese calligraphic images inevitably contain some special noise, such ...
Low Contrast Chinese Rubbing Image Segmentation Based on Gradient Filtering
Intelligent Computing Theories and ApplicationAbstractFor more than 1,500 years, rubbing is one of the most perhaps the oldest of the techniques that have been used in printmaking. Despite the fact that image binarization has been a widely studied in the past decades, segmenting Chinese characters ...
Image denoising with complex ridgelets
In this paper, we propose a novel image denoising method by incorporating the dual-tree complex wavelets into the ordinary ridgelet transform. The approximate shift invariant property of the dual-tree complex wavelet and the high directional sensitivity ...
Comments