Papers by Din-Chang Tseng
Journal of Imaging Science and Technology
Figure 1. The physical dot gain: (a) ideal dots, (b) round dots cause dot gain.
Stroke-based thinning algorithm for Chinese characters
Proceedings of the National Science Council, Republic of China, Part A: Physical Science and Engineering, Jul 1, 1996
Hiding data in triangle meshes by rearranging representation order
International Journal of Innovative Computing Information and Control, Jun 1, 2011
International Journal of Future Computer and Communication, 2018
Hand gesture recognition (HGR) in real-time and with precision has become an important research t... more Hand gesture recognition (HGR) in real-time and with precision has become an important research topic. In this article, a loose hand gesture recognition (LHGR) system based on relational features using a depth sensor is implemented, which not only maintains an impressive accuracy in real-time processing but also enables the user to use loose gestures. HGR can usually be divided into three stages: hand detection, hand feature extraction, and gesture classification. However, the method we propose has been useful in improving all the stages of HGR. In the hand detection stage, we propose a ROI dynamic estimation method and a wrist-cutting method that conform to the characteristics of a human hand. In the feature extraction stage, we use the more reliable relational features which are constructed by local features, global features, and depth coding. In the gesture classification stage, we use three layers of classifiers including finger counting, finger name matching, and coding comparison; these layers are used to classify 16 kinds of hand gestures. In the end, the final output is adjusted by an adaptive decision. The average processing speed per frame is 38.6 ms. Using our method has resulted in an average accuracy of standard gestures of about 98.29%, and an average accuracy of loose gestures of about 88.32%. In summary, our LHGR system can robustly classify hand gestures and still achieve acceptable results for loose gestures.
Remote-sensing image recognition based on wavelet transform and Hausdorff distance
IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477)
Approaches of image enhancement, edge extraction, and line-based matching for remote sensing imag... more Approaches of image enhancement, edge extraction, and line-based matching for remote sensing images are proposed. In the image enhancement, we propose a wave let shrinkage technique to blur the urban regions (i.e. the small-scale texture regions) while preserving the sharpness of large-scale edges (such as highways and rivers) based on a Teager energy criterion. The edge extraction contains wavelet-based edge detection and tracking. The contextual-filter edge detector generates multiresolution gradient images, and then the multiscale edge tracker refines the results as well as reduces the influence of fragment edges and the broken edges. Each extracted edge segment is represented by the coordinates of its mid-point, the logarithm of its length, and its orientation. Then, the matching algorithm based on the Hausdorff distance is applied twice on the two sets of feature vectors for invariant matching.
Medical Image Segmentation Based on the Bayesian Level Set Method
Lecture Notes in Computer Science
A level set method based on the Bayesian risk is proposed for medical image segmentation. At firs... more A level set method based on the Bayesian risk is proposed for medical image segmentation. At first, the image segmentation is formulated as a classification of pixels. Then the Bayesian risk is formed by false-positive and false-negative fractions in a hypothesis test. Through minimizing the average risk of decision in favor of the hypotheses, the level set evolution functional is deduced for finding the boundaries of targets. To prevent the propagating curves from generating excessively irregular shapes and lots of small regions, curvature and gradient of edges in the image are integrated into the functional. Finally, the Euler-Lagrange formula is used to find the iterative level set equation from the derived functional. Comparing with other level-set methods, the proposed approach relies on the optimum decision of pixel classification; thus the approach has more reliability in theory and practice. Experiments show that the proposed approach can accurately extract the complicated shape of targets and is robust for various types of images including high-noisy and low-contrast images, CT, MRI, and ultrasound images; moreover, the algorithm is extendable for multiphase segmentation.
A hybrid physical deformation modeling for laparoscopic surgery simulation
Proceedings of the 22nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (Cat. No.00CH37143), 2000
A hybrid physical-based deformation modeling (HPDM) technique for the authors' laparoscop... more A hybrid physical-based deformation modeling (HPDM) technique for the authors' laparoscopic surgery simulation system is proposed. The proposed technique consists of three components: (i) approximate continuum free-form deformation modeling, (ii) efficient collision detection for non-rigid objects, and (iii) mass-spring modeling for force feedback implementation. Three physical-based deformation modeling techniques which have been applied to surgical simulation are the mass-spring model,
Wavelet-based image denoising using contextual hidden Markov tree model
Journal of Electronic Imaging, 2005
[Journal of Electronic Imaging 14, 033005 (2005)]. Din-Chang Tseng, Ming-Yu Shih. Abstract. ... T... more [Journal of Electronic Imaging 14, 033005 (2005)]. Din-Chang Tseng, Ming-Yu Shih. Abstract. ... The Lena, Baboon, Barbara, Jet, and Peppers images are in size; the Building, Compacted-candy, and Loose-candy images are in size. ...
IEEE Transactions on Robotics and Automation, 1991
A laser-based vision system for computing the location and orientation of 3-D polyhedral surfaces... more A laser-based vision system for computing the location and orientation of 3-D polyhedral surfaces is proposed. In this system, an expanded laser beam passes through a code plate marked with equally spaced vertical and horizontal lines and impinges on a polyhedral object to create a spatial-encoded image for analysis. Then, based on the vanishing points or the directly available line directions of the perceived grid lines on the polyhedral surface, the polyhedral surface orientation can be inferred. In the meantime, the given dimensions of the grid pattern on the plate are used to estimate the depth information of the polyhedral surfaces. More importantly, we shall solve the noise problem that occurs in the real image by a least squares estimation method and an iterative refinement method based on a geometric constraint criterion. Experiments are conducted to provide practical insight into the method. The experimental results indicate that the method is remarkably accurate and stable.
Appearance-preserving view-dependent multiresolution terrain modeling
IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477)
Using the multiresolution technique is one way to archive the real-time rendering for the large-s... more Using the multiresolution technique is one way to archive the real-time rendering for the large-scale environment visualization; however, most multiresolution techniques did not consider the appearance of the environment, then the shapes of the models and environment are degraded during lower-resolution rendering. We present a multiresolution modeling algorithm using quadric error metrics and provide the error metrics with the appearance attributes of a model such as color, normal and texture coordinates to rapidly obtain multiresolution models with high qualities. The view-dependent technique is one way to further improve the rendering performance of multiresolution models. We also provide the view-dependent technique for automatically deciding the proper resolution and model structure based on the view parameters to archive a real-time visualization.
Color segmentation using perceptual attributes
Proceedings., 11th IAPR International Conference on Pattern Recognition. Vol. IV. Conference D: Architectures for Vision and Pattern Recognition,
A general approach for achieving color image segmentation using uniform-chromaticity-scale percep... more A general approach for achieving color image segmentation using uniform-chromaticity-scale perceptual color attributes is proposed. At first chromatic and achromatic areas in a perceptual IHS color space are defined. Then the image is separated into chromatic and achromatic regions according to the region locations in the color space. 1-D histogram thresholding for each color attribute is performed to split the chromatic and achromatic regions, respectively. Finally the region growing is used to solve the oversegmentation problem. In an experiment the power of the proposed approach is demonstrated
Vision-based parking guidance with adaptive isometric transformation
Loose Hand Gesture Recognition Using CNN
Advances in 3D Image and Graphics Representation, Analysis, Computing and Information Technology, 2020
A precise hand gesture recognition (HGR) system is an important facility for human-computer inter... more A precise hand gesture recognition (HGR) system is an important facility for human-computer interaction (HCI). In this paper, we propose a multi-resolution convolutional neural network (CNN) to recognize the loose hand gesture, where loose means that the gestures can be more varied on the bending degrees of fingers, on the direction of palm, and on the bending angles of wrist.The proposed loose hand gesture recognition (LHGR) system learn the low-level features from both color and depth images and then concatenate the low-level features to learn the RGBD (RGB color and Depth) high-level features. The advantage is that it not only suppresses the problem of the inaccurate alignment pixels between color images and deep images, but also reduce the parameters of the CNN model. In addition, we use multi-resolution features to classify the hand gestures; therefore, the proposed model has stronger ability for smaller, farther, and blurrier images. In the training stage, we trained the proposed CNN model using various loose hand gestures to make the CNN more robust. In the experiments, we compared the proposed CNN model in several different architectures; the mAP (mean average precision) is highly to 0.9973. The proposed method has reliability in the scaling and rotation of hand gestures.
International Journal of Computer and Electrical Engineering, 2017
The view scope of a single camera is finite and limited by scene structures. Multi-fisheye camera... more The view scope of a single camera is finite and limited by scene structures. Multi-fisheye cameras can monitor a wide area and trace a complete trajectory of a moving object. In this study, an automatic detection and tracking system with two fisheye cameras for environment surveillance is proposed. The proposed system is composed of two major modules: foreground detection and foreground tracking. The background subtraction method is first applied to extract targets. Then use Kalman filtering for pedestrian motion prediction. A transform table is pre-established to associate multi-cameras data in the overlapping areas. When object across disjoint camera views, the data in the lookup table can provide enough information to realize the moving object in camera views actually belonging to the same object, and keep consistent labels on the object. To improve the reliability of the tracking performance, motion and color appearance features are used to match the detected objects in different cameras. It demonstrates that the proposed method can work well under challenging conditions, such as light change, shadow interference, object occlusion.
International Journal of Hybrid Intelligent Systems, 2005
In this paper, a hybrid approach, which is based on Gaussian smoothing and a genetic algorithm (G... more In this paper, a hybrid approach, which is based on Gaussian smoothing and a genetic algorithm (GA), is proposed for automatic multilevel image thresholding. Using a mixture probability density function of several Gaussian functions to fit an image histogram and then find the optimal threshold(s) is a well-known optimal thresholding method. In the proposed approach, the Gaussian kernel smoothing is used to estimate the number of classes in an image. Since the parameter estimation in the method is typically a nonlinear optimization problem, the parameters used in the mixture of Gaussian functions that give the best fit to the processed histogram are determined using GA. In experiments, synthetic data and real images were processed to evaluate the thresholding performance. The experimental results to confirm the proposed approach are also included.
International Journal of Machine Learning and Computing, 2015
A multiscale texture segmentation approach based on contextual hidden Markov tree (CHMT) model an... more A multiscale texture segmentation approach based on contextual hidden Markov tree (CHMT) model and boundary refinement is proposed. A hidden Markov tree (HMT) model is a probabilistic model for capturing persistence properties of wavelet coefficients without considering clustering properties. We have proposed the CHMT model to enhance the clustering properties by adding extended coefficients associated with wavelet coefficients in every scale. In this study, we train the CHMT parameters for every texture and then use them to compute maximum likelihoods for every dyadic square region at every scale in an image which will be segmented. Then the boundary refinement algorithm is adopted to fuse the different-scale segmented results to improve the final results. We demonstrate the performance of the proposed method on synthetic and aerial images; moreover, the comparison with other methods is also provided to show the effectiveness of the proposed method.
Speckle reduction for remote-sensing images using contextual hidden Markov tree model
IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477)
We propose a contextual hidden Markov tree (CHMT) model by adding intrascale dependences in the h... more We propose a contextual hidden Markov tree (CHMT) model by adding intrascale dependences in the hidden Markov tree (HMT) model to capture more wavelet clustering property and apply the model for SAR image despeckling. Instead of directly adding the transition probabilities between two adjacent hidden states in the HMT model, we add transition probabilities between hidden states of a wavelet
Overlapped-character separation and reconstruction for table-form documents
Proceedings of 3rd IEEE International Conference on Image Processing
We detect the character strokes overlapped by table lines and then reconstruct the broken strokes... more We detect the character strokes overlapped by table lines and then reconstruct the broken strokes after character/line separation. The work is a part of our proposed table-form document analysis system. Tracking models are firstly defined according to the skew angle of a table form to find overlapped strokes. Cut-points are found and paired using a 3×3 mask along tracking models.
Ring Data for Invariant Recognition of Handwwritten Chinese Characters
Journal of Information Science and Engineering, 1998
Fuzzy ring data for invariant handwritten Chinese character recognition
Proceedings of 13th International Conference on Pattern Recognition, 1996
... cognition Din-Chang Tseng and Hung-Pin Chiu ... References [l] DE Rumelhart, GE Hinton, and R... more ... cognition Din-Chang Tseng and Hung-Pin Chiu ... References [l] DE Rumelhart, GE Hinton, and R. J. Williams, "Learning internal representations by error propagation," in Parallel Distributed Processing, MIT Press, Cambridge, [2 ...
Uploads
Papers by Din-Chang Tseng