• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.23 no.4
• /
• pp.805-814
• /
• 1998

An efficient image processing algorithm is proposed to recognize both the height and width modulated barcodes which are rotated and printed at an arbitrary position. The main feature of this algorithm is to utilize the gradient information of a rotated barcode with a Sobel operator. The barcode area is extracted using the gradient information, and the barcode is decoded from the binary image of the extracted area. Theis algorithm is successfully applied to the 4 state and width modulated barcodes. It takes 0.86 secoden to process a letter, and the recognition rate reaches above 98% under various testing conditions. Since both the width and height modulated barcodes are processed with the proposed algorithm, it can be applied to postal service automation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.12
• /
• pp.4326-4344
• /
• 2021

The SinGAN is one of generative adversarial networks that can be trained on a single nature image. It has poor ability to learn more global features from nature image, and losses much local detail information when it generates arbitrary size image sample. To solve the problem, a non-linear function is firstly proposed to control downsampling ratio that is ratio between the size of current image and the size of next downsampled image, to increase the ratio with increase of the number of downsampling. This makes the low-resolution images obtained by downsampling have higher proportion in all downsampled images. The low-resolution images usually contain much global information. Therefore, it can help the model to learn more global feature information from downsampled images. Secondly, the attention mechanism is introduced to the generative network to increase the weight of effective image information. This can make the network learn more local details. Besides, in order to make the output image more natural, the TVLoss function is introduced to the loss function of SinGAN, to reduce the difference between adjacent pixels and smear phenomenon for the output image. A large number of experimental results show that our proposed model has better performance than other methods in generating random samples with fixed size and arbitrary size, image harmonization and editing.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.381-388
• /
• 2022

In instance segmentation, Mask-RCNN is mostly used as a base model. Increasing the performance of Mask-RCNN is meaningful because it affects the performance of the derived model. Mask-RCNN has a transform module for unifying size of input images. In this paper, to improve the Mask-RCNN, we apply deep-learning-based ASSR to the resizing part in the transform module and inject calculated scale information into the model using IM(Integration Module). The proposed IM improves instance segmentation performance by 2.5 AP higher than Mask-RCNN in the COCO dataset, and in the periment for optimizing the IM location, the best performance was shown when it was located in the 'Top' before FPN and backbone were combined. Therefore, the proposed method can improve the performance of models using Mask-RCNN as a base model.

• Journal of the Optical Society of Korea
• /
• v.20 no.1
• /
• pp.107-117
• /
• 2016

We develop a new method to globally calibrate the feature points that are derived from the binocular systems at different positions. A three-DOF (degree of freedom) global calibration system is established to move and rotate the 3D calibration board to an arbitrary position. A three-DOF global calibration model is constructed for the binocular systems at different positions. The three-DOF calibration model unifies the 3D coordinates of the feature points from different binocular systems into a unique world coordinate system that is determined by the initial position of the calibration board. Experiments are conducted on the binocular systems at the coaxial and diagonal positions. The experimental root-mean-square errors between the true and reconstructed 3D coordinates of the feature points are 0.573 mm, 0.520 mm and 0.528 mm at the coaxial positions. The experimental root-mean-square errors between the true and reconstructed 3D coordinates of the feature points are 0.495 mm, 0.556 mm and 0.627 mm at the diagonal positions. This method provides a global and accurate calibration to unity the measurement points of different binocular vision systems into the same world coordinate system.

• Korean Journal of Computational Design and Engineering
• /
• v.9 no.2
• /
• pp.122-132
• /
• 2004

Progressive mesh representation and generation have become one of the most important issues in network-based computer graphics. However, current researches are mostly focused on triangular mesh models. On the other hand, solid models are widely used in industry and are applied to advanced applications such as product design and virtual assembly. Moreover, as the demand to share and transmit these solid models over the network is emerging, the generation and the transmission of progressive solid models depending on specific engineering needs and purpose are essential. In this paper, we present a Cellular Topology-based approach to generating and transmitting progressive solid models from a feature-based solid model for internet-based design and collaboration. The proposed approach introduces a new scheme for storing and transmitting solid models over the network. The Cellular Topology (CT) approach makes it possible to effectively generate progressive solid models and to efficiently transmit the models over the network with compact model size. Thus, an arbitrary solid model SM designed by a set of design features is stored as a much coarser solid model SM/sup 0/ together with a sequence of n detail records that indicate how to incrementally refine SM/sup 0/ exactly back into the original solid model SM = SM/sup 0/.

• Journal of Korea Multimedia Society
• /
• v.19 no.8
• /
• pp.1310-1319
• /
• 2016

Face detection is the first step in a wide range of face applications. However, detecting faces in the wild is still a challenging task due to the wide range of variations in pose, scale, and occlusions. Recently, many deep learning methods have been proposed for face detection. However, further improvements are required in the wild. Another important issue to be considered in the face detection is the computational complexity. Current state-of-the-art deep learning methods require a large number of patches to deal with varying scales and the arbitrary image sizes, which result in an increased computational complexity. To reduce the complexity while achieving better detection accuracy, we propose a fully convolutional network-based face detection that can take arbitrarily-sized input and produce feature maps (heat maps) corresponding to the input image size. To deal with the various face scales, a multi-scale network architecture that utilizes the facial components when learning the feature maps is proposed. On top of it, we design multi-task learning technique to improve detection performance. Extensive experiments have been conducted on the FDDB dataset. The experimental results show that the proposed method outperforms state-of-the-art methods with the accuracy of 82.33% at 517 false alarms, while improving computational efficiency significantly.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.05a
• /
• pp.165-170
• /
• 2003

This paper presents an integrated approach to unconstrained face recognition in arbitrary scenes. The front end of the system comprises of a scale and pose tolerant face detector. Scale normalization is achieved through novel combination of a skin color segmentation and log-polar mapping procedure. Principal component analysis is used with the multi-view approach proposed in[10] to handle the pose variations. For a given color input image, the detector encloses a face in a complex scene within a circular boundary and indicates the position of the nose. Next, for recognition, a radial grid mapping centered on the nose yields a feature vector within the circular boundary. As the width of the color segmented region provides an estimated size for the face, the extracted feature vector is scale normalized by the estimated size. The feature vector is input to a trained neural network classifier for face identification. The system was evaluated using a database of 20 person's faces with varying scale and pose obtained on different complex backgrounds. The performance of the face recognizer was also quite good except for sensitivity to small scale face images. The integrated system achieved average recognition rates of 87% to 92%.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.6
• /
• pp.1312-1317
• /
• 2003

This paper presents an integrated approach to unconstrained face recognition in arbitrary scenes. The front end of the system comprises of a scale and pose tolerant face detector. Scale normalization is achieved through novel combination of a skin color segmentation and log-polar mapping procedure. Principal component analysis is used with the multi-view approach proposed in[10] to handle the pose variations. For a given color input image, the detector encloses a face in a complex scene within a circular boundary and indicates the position of the nose. Next, for recognition, a radial grid mapping centered on the nose yields a feature vector within the circular boundary. As the width of the color segmented region provides an estimated size for the face, the extracted feature vector is scale normalized by the estimated size. The feature vector is input to a trained neural network classifier for face identification. The system was evaluated using a database of 20 person's faces with varying scale and pose obtained on different complex backgrounds. The performance of the face recognizer was also quite good except for sensitivity to small scale face images. The integrated system achieved average recognition rates of 87% to 92%.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.24 no.4
• /
• pp.382-391
• /
• 2021

Unsupervised domain adaptation often gives impressive solutions to handle domain shift of data. Most of current approaches assume that unlabeled target data to train is abundant. This assumption is not always true in practices. To tackle this issue, we propose a general solution to solve the domain gap minimization problem without any target data. Our method consists of two regularization steps. The first step is a pixel regularization by arbitrary style transfer. Recently, some methods bring style transfer algorithms to domain adaptation and domain generalization process. They use style transfer algorithms to remove texture bias in source domain data. We also use style transfer algorithms for removing texture bias, but our method depends on neither domain adaptation nor domain generalization paradigm. The second regularization step is a feature regularization by feature alignment. Adding a feature alignment loss term to the model loss, the model learns domain invariant representation more efficiently. We evaluate our regularization methods from several experiments both on small dataset and large dataset. From the experiments, we show that our model can learn domain invariant representation as much as unsupervised domain adaptation methods.

• Proceedings of the IEEK Conference
• /
• 2002.07a
• /
• pp.149-151
• /
• 2002

We had been developing the variety of parallel processing systems in order to improve the processing speed of visual media applications. These systems were using multi-access memory system(MAMS) as a parallel memory system, which provides the capability of the simultaneous accesses of image points in a line-segment with an arbitrary degree, which is required in many low-level image processing operations such as edge or line detection in a particular direction, and so on. But, the performance of these systems did not give a faithful speed because of asynchronous feature between MAMS and processing elements. To improve the processing speed of these systems, we have been investigated a pipelined parallel processing system using MAMS. Although the system is considered as being the single instruction multiple data(SIMD) type like the early developed systems, the performance of the system yielded about 2.5 times faster speed.