Journal Articles by Nguyen Tran Lan Anh
Fingerprint Authentification System in Banking Transaction
Conference Proceedings by Nguyen Tran Lan Anh
Proceedings of the 7th International Conference on Ubiquitous Information Management and Communication (ICUIMC), Jan 2013
In this paper, a novel level set approach is presented to segment only indicated objects from nat... more In this paper, a novel level set approach is presented to segment only indicated objects from natural color images. To solve this challenging problem, users first initialize a contour inside the concerned object. The proposed level set model is then applied to achieve the object containing this initial region regardless of its shape and size. In our novel framework, the energy functional is a combination of two energy terms based on Bhattacharya flow and graph partitioning active contour. So that movement of the contour can be controlled toward object boundaries by correlative relationships between the interested region and surroundings. The experimental results obtained from our natural image collection show that the suggested method yields more accurate and better performance when comparing to other segmentation algorithms.
Proceedings of the 21st International Conference on Pattern Recognition (ICPR), Nov 2012
Segmenting objects from color images to obtain useful information is a challenging research area ... more Segmenting objects from color images to obtain useful information is a challenging research area recently. In this paper, a novel algorithm by combining a saliency map with an extension of a geometric active contour model is proposed to automatically segment the object of interest. The saliency map is first generated from the input image by a histogram based contrast method. The most salient regions are then detected as dominant parts of the object. After that, a contour is initialized using salient regions determined. Finally, by applying a geometric active contour model, the contour starts evolving iteratively to segment object boundaries. Experimental results attained on various natural scene images have shown that our proposed method is able to not only replace manual initialized contour and improve the accuracy, noise robustness of segmentation but converge to an optimal solution earlier than recent active contour models as well.
Proceedings of the 6th International Conference on Ubiquitous Information Management and Communication (ICUIMC), Feb 2012
In this paper, we propose a novel segmentation algorithm for color images. This method is a combi... more In this paper, we propose a novel segmentation algorithm for color images. This method is a combination of edge information with region information and a geometric active contour without re-initialization, called distance regularized level set evolution. The information given by a new edge detector using morphological gradient is more accurate than normal gradient computing methods for color images. And the information of the region containing objects is relied on Chan-Vese minimal variance criterion. With both of these information, the model can have its initial contour that is more flexible to construct anywhere, fast to evolve and quite exact to stop at the boundary of objects. The suggested algorithm has been applied on natural color images with good performance. Some experimental results have shown to compare our model with others with respect to accuracy and computational efficiency.
Proceedings of The 3rd International Conference on Internet (ICONI), Dec 2011
Segmenting objects in images is an important problem of image processing and computer vision. The... more Segmenting objects in images is an important problem of image processing and computer vision. There have been various level set methods applied on gray scale images. However, in practice, these methods will give negative effects to color images such as the distortion when using its gray-scale gradient information. In this paper, object segmentation based on a modification of a new type of level set method named distance regularized level set evolution (DRLSE) in color images is proposed. The speed function here will be designed to use color gradient information of the image. Finally, experimental results are provided to show the accuracy of this method when applying to different color images.
The Conference on Human-Computer Interaction (HCI), Jan 2011
AbstractˇˇIn this paper, a motion tracking method for built-in camera mobile phones is proposed b... more AbstractˇˇIn this paper, a motion tracking method for built-in camera mobile phones is proposed based on global motion estimation. By using Three-Step Search algorithm, block based motion vectors are first extracted from each of two adjacent frames. Motion vector outliers are then removed by a rejection cascade algorithm. Finally, a perspective model is applied to estimate the global motion vector and perform the movement of mobile phone. The simulated experimental results from real recording videos show that the proposed method generates acceptable motion estimation by performing a smooth trajectory under most conditions of movement in a reasonable processing time.
Uploads
Journal Articles by Nguyen Tran Lan Anh
Conference Proceedings by Nguyen Tran Lan Anh