• Journal of Korea Multimedia Society
• /
• v.12 no.7
• /
• pp.929-940
• /
• 2009

This paper presents multipurpose watermarking scheme in coutourlet transform domain for copyright protection, authentication and transform detection. Since contourlet transform can detect more multi direction edge and smooth contour than wavelet transform, the proposed scheme embeds multi watermarks in contourlet domain based on 4-level Laplacian pyramid and 2-level directional filter bank. In the first stage of the robust watermarking scheme for copyright protection, we generates the sequence of circle patterns according to watermark bits and projects these patterns into the average of magnitude coefficients of high frequency directional subbands. Then the watermark bit is embedded into variance distribution of the projected magnitude coefficients. In the second stage that is the semi-fragile watermarking scheme for authentication and transform detection, we embed the binary watermark image in the low frequency subband of higher level by using adaptive quantization modulation scheme. From the evaluation experiment using Checkmark 2.1, we verified that the proposed scheme is superior to the conventional scheme in a view of the robustness and the invisibility.

• Journal of Biomedical Engineering Research
• /
• v.41 no.4
• /
• pp.147-153
• /
• 2020

This study was to evaluate the usability of self-developed phantom for evaluating automatic exposure control (AEC) using three-dimensions (3D) printer. 3D printer of fused deposition modeling (FDM) type was utilized to make the self-developed AEC phantom and image acquisitions were conducted by two different type of scanners. The self-developed AEC phantom consisted of four different size of portions. As a result, two types of phantom (pyramid and pentagon shape) were created according to the combination of the layers. For evaluating the radiation dose with the two types of phantom, the values of tube current, computed tomography dose index volume (CTDIvol), and dose length product (DLP) were compared. As a result, it was confirmed that the values of tube current were properly reflected according to the thickness, and the CTDIvol and DLP were not significantly changed regardless of AEC functions of different scanners. In conclusion, the self-developed phantom by using 3D printer could assess whether the AEC function works well. So, we confirmed the possibility that a self-made phantom could replace the commercially expensive AEC performance evaluation phantom.

• Journal of the Institute of Convergence Signal Processing
• /
• v.13 no.1
• /
• pp.12-18
• /
• 2012

Stereo correspondence is the central problem of stereo vision. In this paper, we propose a stereo matching scheme based on a genetic algorithm using a multi-resolution method and AD-Census. The proposed approach considers the matching environment as an optimization problem and finds the disparity by using a genetic algorithm And adaptive chronosome structure using edge pixels and crossover mechanism are employed in this technique. A cost function is composes of certain constraints whice are commonly used in stereo matching. AD-Census measure is applied to reduce disparity error. To increase the efficiency of process, we apply image pyramid method to stereo matching and calculate the initial disparity map at the coarsest resolution. Then initial disparity map is propagated to the next finer resolution, interpolated and performed disparity refinement using local feature vector. We valid our method not only reduces the search time for correspondence compared with conventional GA-based method but also ensures the validity of matching.

• Journal of Institute of Control, Robotics and Systems
• /
• v.8 no.1
• /
• pp.21-27
• /
• 2002

Inspection and shape measurement of three-dimensional objects are widely needed in industries for quality monitoring and control. A number of visual or optical technologies have been successfully applied to measure three-dimensional surfaces. However, those conventional visual or optical methods have inherent shortcomings such as occlusion and variant surface reflection. X-ray vision system can be a good solution to these conventional problems, since we can extract the volume information including both the surface geometry and the inner structure of any objects. In the x-ray system, the surface condition of an object, whether it is lambertian or specular, does not affect the inherent characteristics of its x-ray images. In this paper, we propose a three-dimensional x-ray imaging method to reconstruct a three dimensional structure of an object out of two dimensional x-ray image sets. To achieve this by the proposed method, two or more x-ray images projected from different views are needed. Once these images are acquired, the simultaneous algebraic reconstruction technique(SART) is usually utilized. Since the existing SART algorithms have several shortcomings such as low performance in convergence and different convergence within the reconstruction volume of interest, an advanced SART algorithm named as USART(uniform SART) is proposed to avoid such shortcomings and improve the reconstruction performance. Because, each voxel within the volume is equally weighted to update instantaneous value of its internal density, it can achieve uniform convergence property of the reconstructed volume. The algorithm is simulated on various shapes of objects such as a pyramid, a hemisphere and a BGA model. Based on simulation results the performance of the proposed method is compared with that of the conventional SART method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.10
• /
• pp.5179-5196
• /
• 2019

To explore an effective non-invasion medical imaging diagnostics approach for hepatocellular carcinoma (HCC), we propose a method based on adopting the multiple technologies with the multi-parametric data fusion, transfer learning, and multi-scale deep feature extraction. Firstly, to make full use of complementary and enhancing the contribution of different modalities viz. multi-parametric MRI images in the lesion diagnosis, we propose a data-level fusion strategy. Secondly, based on the fusion data as the input, the multi-scale residual neural network with SPP (Spatial Pyramid Pooling) is utilized for the discriminative feature representation learning. Thirdly, to mitigate the impact of the lack of training samples, we do the pre-training of the proposed multi-scale residual neural network model on the natural image dataset and the fine-tuning with the chosen multi-parametric MRI images as complementary data. The comparative experiment results on the dataset from the clinical cases show that our proposed approach by employing the multiple strategies achieves the highest accuracy of 0.847±0.023 in the classification problem on the HCC differentiation. In the problem of discriminating the HCC lesion from the non-tumor area, we achieve a good performance with accuracy, sensitivity, specificity and AUC (area under the ROC curve) being 0.981±0.002, 0.981±0.002, 0.991±0.007 and 0.999±0.0008, respectively.

• Journal of Electrical Engineering and Technology
• /
• v.9 no.2
• /
• pp.739-748
• /
• 2014

In this paper, we propose pyramid appearance and global structure action descriptors on both RGB and depth motion history images and a model-free method for human action recognition. In proposed algorithm, we firstly construct motion history image for both RGB and depth channels, at the same time, depth information is employed to filter RGB information, after that, different action descriptors are extracted from depth and RGB MHIs to represent these actions, and then multimodality information collaborative representation and recognition model, in which multi-modality information are put into object function naturally, and information fusion and action recognition also be done together, is proposed to classify human actions. To demonstrate the superiority of the proposed method, we evaluate it on MSR Action3D and DHA datasets, the well-known dataset for human action recognition. Large scale experiment shows our descriptors are robust, stable and efficient, when comparing with the-state-of-the-art algorithms, the performances of our descriptors are better than that of them, further, the performance of combined descriptors is much better than just using sole descriptor. What is more, our proposed model outperforms the state-of-the-art methods on both MSR Action3D and DHA datasets.

• Progress in Medical Physics
• /
• v.25 no.3
• /
• pp.128-138
• /
• 2014

In gated radiation therapy (gRT), due to residual motion, beam delivery is intended to irradiate not only the true extent of disease, but also neighboring normal tissues. It is desired that the delivery covers the true extent (i.e. clinical target volume or CTV) as a minimum, although target moves under dose delivery. The objectives of our study are to validate if the intended dose is surely delivered to the true target in gRT and to quantitatively understand the trend of dose delivery on it and neighboring normal tissues when gating window (GW), motion amplitude (MA), and CTV size changes. To fulfill the objectives, experimental and computational studies have been designed and performed. A custom-made phantom with rectangle- and pyramid-shaped targets (CTVs) on a moving platform was scanned for four-dimensional imaging. Various GWs were selected and image integration was performed to generate targets (internal target volume or ITV) for planning that included the CTVs and internal margins (IM). The planning was done conventionally for the rectangle target and IMRT optimization was done for the pyramid target. Dose evaluation was then performed on a diode array aligned perpendicularly to the gated beams through measurements and computational modeling of dose delivery under motion. This study has quantitatively demonstrated and analytically interpreted the impact of residual motion including penumbral broadening for both targets, perturbed but secured dose coverage on the CTV, and significant doses delivered in the neighboring normal tissues. Dose volume histogram analyses also demonstrated and interpreted the trend of dose coverage: for ITV, it increased as GW or MA decreased or CTV size increased; for IM, it increased as GW or MA decreased; for the neighboring normal tissue, opposite trend to that of IM was observed. This study has provided a clear understanding on the impact of the residual motion and proved that if breathing is reproducible gRT is secure despite discontinuous delivery and target motion. The procedures and computational model can be used for commissioning, routine quality assurance, and patient-specific validation of gRT. More work needs to be done for patient-specific dose reconstruction on CT images.

• The Journal of Art Theory & Practice
• /
• no.7
• /
• pp.165-188
• /
• 2009

In the development of linear perspective, Brook Taylor's theory has achieved a special position. With his method described in Linear Perspective(1715) and New Principles of Linear Perspective(1719), the subject of linear perspective became a generalized and abstract theory rather than a practical method for painters. He is known to be the first who used the term 'vanishing point'. Although a similar concept has been used form the early stage of Renaissance linear perspective, he developed a new method of British perspective technique of measure points based on the concept of 'vanishing points'. In the 15th and 16th century linear perspective, pictorial space is considered as independent space detached from the outer world. Albertian method of linear perspective is to construct a pavement on the picture in accordance with the centric point where the centric ray of the visual pyramid strikes the picture plane. Comparison to this traditional method, Taylor established the concent of a vanishing point (and a vanishing line), namely, the point (and the line) where a line (and a plane) through the eye point parallel to the considered line (and the plane) meets the picture plane. In the traditional situation like in Albertian method, the picture plane was assumed to be vertical and the center of the picture usually corresponded with the vanishing point. On the other hand, Taylor emphasized the role of vanishing points, and as a result, his method entered the domain of projective geometry rather than Euclidean geometry. For Taylor's theory was highly abstract and difficult to apply for the practitioners, there appeared many perspective treatises based on his theory in England since 1740s. Joshua Kirby's Dr. Brook Taylor's Method of Perspective Made Easy, Both in Theory and Practice(1754) was one of the most popular treatises among these posterior writings. As a well-known painter of the 18th century English society and perspective professor of the St. Martin's Lane Academy, Kirby tried to bridge the gap between the practice of the artists and the mathematical theory of Taylor. Trying to ease the common readers into Taylor's method, Kirby somehow abbreviated and even omitted several crucial parts of Taylor's ideas, especially concerning to the inverse problems of perspective projection. Taylor's theory and Kirby's handbook reveal us that the development of linear perspective in European society entered a transitional phase in the 18th century. In the European tradition, linear perspective means a representational system to indicated the three-dimensional nature of space and the image of objects on the two-dimensional surface, using the central projection method. However, Taylor and following scholars converted linear perspective as a complete mathematical and abstract theory. Such a development was also due to concern and interest of contemporary artists toward new visions of infinite space and kaleidoscopic phenomena of visual perception.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.10
• /
• pp.35-45
• /
• 2024

In this paper, we propose Multi-Scale Feature Knowledge Distillation for 3D Object Detection (M3KD), which extracting knowledge from the teacher model, and transfer to the student model consider with multi-scale feature map. To achieve this, we minimize L2 loss between feature maps at each pyramid level of the student model with the correspond teacher model so student model can mimic the teacher model backbone information which improves the overall accuracy of the student model. We apply the class logits knowledge distillation used in the image classification task, by allowing student model mimic the classification logits of the teacher model, to guide the student model to improve the detection accuracy. In KITTI (Karlsruhe Institute of Technology and Toyota Technological Institute) dataset, our M3KD (Multi-Scale Feature Knowledge Distillation for 3D Object Detection) student model achieves 30% inference speed improvement compared to the teacher model. Additionally, our method achieved an average improvement of 1.08% in 3D mean Average Precision (mAP) across all classes and difficulty levels compared to the baseline student model. Furthermore, when integrated with the latest knowledge distillation methods such as PKD and SemCKD, our approach achieved an additional 0.42% and 0.52% improvement in 3D mAP, respectively, further enhancing performance.