Kết quả thực hiện các mục tiêu khoa học và công nghệ năm 2014-2015 tại TTHL (15/03/2016)
 
Kết quả thực hiện các mục tiêu khoa học và công nghệ năm 2014-2015 tại TTHL cụ thể như sau:

TTNội dung Năm công bố/xuất bản/thành lậpĐơn vị chủ trì, tác giảTóm tắt nội dung
1Image Denoising by Addaptive Non-Local Bilatetal Filter 2014International Journal of Computer Applications 99(12):4-10. Published by Foundation of Computer Science, New York, USA. ISBN: 973-93-80883-42-4, pp.4-10, August 2014 / TS. Đào Nam AnhIn fields such as demosaicking, texture removal, dynamic range compression, and photo enhancement many imaging modalities operate with images corrupted by different noise models. Bilateral filter and non-local mean filter are often applied for deduction of noise. This paper presents a new adaptive bilateral filter model to reconstruct edges by choosing neighborhood with non-local mean concept. The method yields considerable gain reduction of noise and keep edges better than original method. Basing in visual inspection, the new method is considered as effective even in case of mixed noise.
2Image Denoising by Two-Pass of Total Variation Filter2014International Journal of Computer Applications 98(17):24-29. Published by Foundation of Computer Science, New York, USA. doi: 10.5120/17276-7706, ISBN: 973-93-80882-77-5, pp.24-29, July 2014 / TS. Đào Nam AnhTotal variation based methods are widely applied for image enhancement and particularly for de-noising. The majority of these is designed for a specific noise model. The alternative total variation based approach proposed here can deal with multiple noise models via two-pass iterative algorithm basing on total variation. The first pass is designed for draft denoising and to detect noise region. The second pass restores the noise region by total variation based inpainting. Experiments on Salt & Pepper, Gaussian, Speckle, Poisson, and Impulse noise models demonstrate the effectiveness of the proposed method.
3An Adaptive Bilateral Filter For Inpainting2014Fourth International Conference on Emerging Applications of Information Technology (EAIT 2014), ISBN 978-1-4799-4272-5/14 (print) IEEE, DOI 10.1109/EAIT.2014.13, pages 237-242, 2014/ TS. Đào Nam AnhAn adaptive model of bilateral filter is presented for digital inpainting. The model works by transforming inpainting into an equivalent energy condition minimization and generation of patches for missing areas by interpolating within working frame. It combines knowledge of local structure by bilateral filter and intensive value. Bilateral filter is adapted to missing regions to check similarity of regions to fill-in. Standard deviation in range kernel of the filter is regulated by total variation. This helps to create patches that keep edges. Total variation is also efficient for detection of missing pixels which possibly stay on edges. Benefit of the model was demonstrated in experiment of inpainting for gray and color images.
4Normalized Cuts Segmentation with Non-Local Mean and Bilateral Filter2014XVII National Conference: Some Selected Issues of Information Technology and Communication, 2014, pp 235-241, NXBKHKT, Hanoi, 2014/ TS. Đào Nam AnhSegmentation for images having boundary with unstable gradient is still significant challenge. Combination of bilateral filter and non local operators is an approach for this segmentation question. In this paper non-local operators with bilateral filter are put in energy functions for unsupervised normalized cuts segmentation. The study yields a non-local bilateral filter segmentation algorithm using both local and non-local comparisons between pairs of graph notes. Experimental results show that the image segmentation approach is very competitive to manage patches and boundary.
5Iterative Bilateral Filter and Non-Local Mean2014International Journal of Computer Applications 106(11):33-38. Published by Foundation of Computer Science, New York, USA. doi:10.5120/18566-9818, ISBN: 973-93-80883-73-5, pp.33-38, November 2014/ TS. Đào Nam AnhUsing local or non-local features has proven to be a competent approach for denoising images. As noise and edges have similar effect of changes in gradient in many cases, noise allocation for denoising is still significant challenge. This work addresses the classic problem but introducing the combination concept of local and non-local factors with deviation refinement procedure. A new algorithm of the concept is proposed to ameliorate noise reduction. Sensitivity of noise detection is examined by iterative non-local mean and bilateral filter with refinement of range deviation. The final methodology is tested with Gaussian noise and compared with both non-local mean, bilateral filter. Experiment demonstrates improvement of denoising level in the new algorithm.
6Local Adaptive Bilateral Filter with Variation for Deblurring.2014International Journal of Computer Applications 86(15):13-18. Published by Foundation of Computer Science, New York, USA. doi: 10.5120/15060-33955, ISBN : 973-93-80879-95-9, pp.13-18, January 2014/ TS. Đào Nam AnhIn this study, alternative application of bilateral filter for image deblurring and enhancement is discovered. The concept of Total variation model is added to BF. Based on analyzing force distribution rules of variance, standard deviation is managed to distinguish degree of degrade. The optimization solution of total variation is gained by tracking minimum change channels and keep maximum edges. Experimentation proves that the new V-ABF can solve the deblurring problem where original BF is solution for de-noising.
7Image Segmentation by Superpixels and Gradients2014Journal of Science and Technology for Energy (JSTE). ISSN: 1859-4557, pages 34-43, No 7 2014/ TS. Đào Nam AnhThis work presents an algorithm using superpixels and local gradients for image segmentation. The algorithm is integrated concept of top-down and bottom-up. Pixels of input image are grouped into superpixels, and then superpixels are merged into bigger segment. Information of gradients in local frame is essential for the merging process. The concept helps to avoid too much using global information in order to get advance in time complexity.
8Divergence Filter for Saliency20157th International Conference On Knowledge And Systems Engineering (KSE), 978-1-4673-8013-3/15 IEEE, DOI 10.1109/KSE.2015.8, pp.238-243, October 2015/ TS. Đào Nam AnhDetection of regions with high visual attention from image has various applications including advertising design where ads are often associated with relevant semantic visual information. The salient regions in the image/video have to be identified in a consistent way, even if original objects or background are texture scene. This is achieved by solving combinatorial problem of down-sampling that searches for the optimal texture region map. The complexity of this solution makes it impractical. The problem becomes easy by a new approach for saliency detection. It is based on the spatial attention model that evaluates divergence of a given local region from its surrounding where objects and background can be texture scene.
9Saliency Guided Interpolation for Super-Resolution2015XVIII National Conference: Some Selected Issues of Information Technology and Communication, 5- 6/11/2015, HCM/ Dao Nam Anh, Nguyen Huu Quynh, Nguyen Hong SonSaliency maps permit new possibilities for estimation of high frequency details which are blurred in images. In this work, an original approach for enhancement of image resolution from a single low resolution image by employing saliency map and adaptive Gaussian filters is reported. Initial details are generated through a spatial divergent Gaussian kernel which fills the upsized image incrementally. Saliency map is established from analysis of distinctness factors. Another kernel on intensity/color is followed to reflect the edges and remove noise. Fine details are delivered by applying particular enhancement in salient regions. Hence a specific fundamental method to generate a higher resolution image for lower resolution images with saliency map and the adaptive Gaussian kernels is presented. The human visual perception is used for quantitative regularization in the interpolation and features of salient regions are improved.
10Delaunay Sparse Saliency20151st International Workshop on Pattern Recognition for Multimedia Content Analysis (PR4MCA), 978-1-4673-8013-3/15 IEEE, DOI 10.1109/KSE.2015.32, pp.383-388, October 2015/ TS. Đào Nam AnhManual identifying region with visual attention in image/video frames is not a facile task. An algorithm which automatically reveals salient region from a single image is presented. The challenging problem of saliency detection is tackled by the Delaunay triangulation with an advanced salient criteria. The criteria addresses to difference of intensive value and image feature which is based on local and global mean. Delaunay mesh points are generated by the feature and then Delaunay triangulation is created. An adaptive sparse bilateral filter is taken advantage of to uncover local saliency and to set up the mesh points. The filter is sparse as it’s applied for the mesh points only. Salient regions are extracted and boundary of the regions are followed by the mesh edges. Both points and edges of the Delaunay mesh are used to solve the saliency problem. Experimental results with various image categories from a saliency benchmark demonstrates reliable performance. Furthermore, it’s very explicative as the Delaunay mesh presents saliency map illustratively.
11Bilateral Filtering with Clusters by Expectation Maximization2015Journal of Science and Technology (JST), ISSN: 1859-0209, Section on information and communication technology, No. 6 (4-2015), pp. 30-39, April 2015/ TS. Đào Nam AnhImage restoration keeping sharp edges is achieved by bilateral filter. In this paper, an approach to improve edges for the filter is presented. The proposed algorithm relies on clustering by Expectation Maximization that produced clusters of intensive values, followed by a stage where standard deviation of Gaussian filters for scales of the spatial and intensity are adjusted by features of the clusters. This makes the filters have adaptive levels of smoothing for specific clusters and helps to preserve edges while remove noise.
Experiments and evaluation by PSNR metrics indicated the restoration quality enhanced and the efficacy of the proposed adaptive bilateral filter algorithm.
12Inverse Bilateral Filter for Saliency2015International Journal of Computer Applications 118(10):11-19, Published by Foundation of Computer Science, New York, USA. DOI: 10.5120/20780-3343, ISBN: 973-93-80886-79-4, pp.11-19, May 2015/ TS. Đào Nam AnhThe analysis and automatic detection of visual salient image regions has been the subject of considerable research useful in object segmentation, adaptive compression and re-targeting. However, the nature of the essential mechanisms intervening human visual saliency remains elusive. To assess the validity of salient regions some kind of prior model is consistently required. This paper proposes a new model using inverse bilateral fitter that allows the system to output saliency maps with salient objects in their context. The filter is described firstly for automatically learning local contrast distribution to accurately predict salient image regions. Along with the contrast distribution checking, local opposition is analyzed by the second application of the inverse bilateral filter to establish fuzzy boundary of salient regions in form of trimap. This approach is shown to increase the reliability of identifying visual salient objects. Output from the research has potential applications in the areas of object detection and recognition.
13Saliency Detection with Voronoi Diagram2015International Journal of Computer Applications 118(12):27-34, Published by Foundation of Computer Science, New York, USA. DOI: 10.5120/20798-3468, ISBN: 973-93-80886-81-8, pp.27-34, May 2015/ Dao Nam Anh, Nguyen Huu QuynhMany applications are serviced by the Voronoi tessellation required to split image into Voronoi regions. An automatic method to learn and detect salient region for color image with support of the Voronoi diagram is presented. Salient regions are modeled as flexible circumstance corresponding to centers of mass. The centers are predicted by local contrast-based representation with local maxima. Results are demonstrated that are very competitive with other recent saliency map detection schemes and show robustness to capture visual attention objects. Our major contributions are the local maxima based method for allocation of Voronoi centroids and the Gaussian-based filter for estimating attention degrees. To show the effectiveness of the approach, saliency maps are detected for images of MSRA saliency object database by some state-of-the-art methods. The strengths and the weaknesses of the approach are considered, with a special focus on the context based salient regions  a challenging task which can be found in wide range of applications addressed in computer vision.
14Smooth Context based Color Transfer2015International Journal of Computer Applications 116(15):29-37, Published by Foundation of Computer Science, New York, USA. DOI: 110.5120/20413-2825, ISBN: 973-93-80886-26-0, pp.29-37, April 2015/ TS. Đào Nam AnhColor transfer is an emerging framework for dealing with ubiquitous color manipulation in media such as documents and images. Despite the notable progress made in the field, there remains a need for designers that can represent the same information in personalization and corresponding to media context. This work presents adaptive color transfer method using cross-disciplinary interaction of semantic context and bilateral filters. Colors in the method are transferred softly in matching with saliency distributed context. Preliminary results show that the framework is highly keeping consistency and promising. Consequently in this work, a solution of tone mapping by color transfer is introduced. Experimental results are further showed pertaining for automatic handling colors and contrast.
15Segmentation by Incremental Clustering2015International Journal of Computer Applications 111(12):23-30, Published by Foundation of Computer Science, New York, USA. DOI: 10.5120/19591-1360, ISBN: 973-93-80885-25-6, pp.23-30, February 2015/ TS. Đào Nam AnhA method for unsupervised segmentation by incremental clustering is introduced. Inspired by incremental approach and correlation clustering, clusters are added and refined during segmentation process. Correlation clustering is to keep away from pre-definition for number of clusters and incremental approach is to avoid re-clustering that is needed in iterative methods. The Gaussian spatial kernel is involved like a part of similarity function to cover local image structure. Cluster representative is updated efficiently to satisfy the old and new similarity constraints rather than re-clustering the entire image. Experimental results are discussed and show that the algorithm requires reasonable computational complexity while gaining a comparable or better segmentation quality than standard methods.






Tin mới
Đăng ký môn học trực tuyến