Training an Active Random Field for Real-Time Image Denoising TeX Export

@article{citeulike:6045093, abstract = {<para> Many computer vision problems can be formulated in a Bayesian framework based on Markov random fields (MRF) or conditional random fields (CRF). Generally, the MRF/CRF model is learned independently of the inference algorithm that is used to obtain the final result. In this paper, we observe considerable gains in speed and accuracy by training the MRF/CRF model together with a fast and suboptimal inference algorithm. An active random field (ARF) is defined as a combination of a MRF/CRF based model and a fast inference algorithm for the MRF/CRF model. This combination is trained through an optimization of a loss function and a training set consisting of pairs of input images and desired outputs. We apply the ARF concept to image denoising, using the Fields of Experts MRF together with a 1–4 iteration gradient descent algorithm for inference. Experimental validation on unseen data shows that the ARF approach obtains an improved benchmark performance as well as a 1000–3000 times speedup compared to the Fields of Experts MRF. Using the ARF approach, image denoising can be performed in real-time, at 8fps on a single CPU for a 256<formula formulatype="inline"><tex Notation="TeX">\$, times,\$</tex></formula>256 image sequence, with close to state-of-the-art accuracy. </para>}, author = {Barbu, Adrian}, citeulike-article-id = {6045093}, citeulike-linkout-0 = {http://dx.doi.org/10.1109/TIP.2009.2028254}, citeulike-linkout-1 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5173526}, day = {24}, doi = {10.1109/TIP.2009.2028254}, journal = {Image Processing, IEEE Transactions on}, keywords = {image\_processing}, month = {July}, number = {11}, pages = {2451--2462}, posted-at = {2009-10-30 21:47:27}, priority = {2}, title = {Training an Active Random Field for Real-Time Image Denoising}, url = {http://dx.doi.org/10.1109/TIP.2009.2028254}, volume = {18}, year = {2009} }

TY - JOUR ID - citeulike:6045093 L3 - citeulike-article-id:6045093 N2 - <para> Many computer vision problems can be formulated in a Bayesian framework based on Markov random fields (MRF) or conditional random fields (CRF). Generally, the MRF/CRF model is learned independently of the inference algorithm that is used to obtain the final result. In this paper, we observe considerable gains in speed and accuracy by training the MRF/CRF model together with a fast and suboptimal inference algorithm. An active random field (ARF) is defined as a combination of a MRF/CRF based model and a fast inference algorithm for the MRF/CRF model. This combination is trained through an optimization of a loss function and a training set consisting of pairs of input images and desired outputs. We apply the ARF concept to image denoising, using the Fields of Experts MRF together with a 1–4 iteration gradient descent algorithm for inference. Experimental validation on unseen data shows that the ARF approach obtains an improved benchmark performance as well as a 1000–3000 times speedup compared to the Fields of Experts MRF. Using the ARF approach, image denoising can be performed in real-time, at 8fps on a single CPU for a 256<formula formulatype="inline"><tex Notation="TeX">$, times,$</tex></formula>256 image sequence, with close to state-of-the-art accuracy. </para> IS - 11 JF - Image Processing, IEEE Transactions on EP - 2462 TI - Training an Active Random Field for Real-Time Image Denoising VL - 18 SP - 2451 KW - image_processing AU - Barbu, Adrian PY - 2009/07/24/ UR - http://dx.doi.org/10.1109/TIP.2009.2028254 ER -