• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.39 no.1
• /
• pp.32-39
• /
• 2002

Vector quantization for image compression needs expensive encoding time to find the closest codeword to the input vector. This paper proposes a search algorithm for fast vector quantization encoding. Firstly, we derive a robust condition based on the efficient topological structure of the codebook to dramatically eliminate unnecessary matching operations from the search procedure. Then, we Propose a fast search algorithm using the elimination condition. Simulation results show that with little preprocessing and memory cost, the encoding time of the proposed algorithm is reduced significantly while the encoding quality remains the same with respect to the full search algorithm. It is also found that the Proposed algorithm outperforms the existing search algorithms.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.04b
• /
• pp.592-594
• /
• 2000

변형모델(deformable model)은 볼륨의료영상(volumetric medical image)으로부터 복잡한 인체기관의 3차원적 경계를 분할해내기 위해 효과적인 방법을 제공한다. 그러나, 기존 변형모델은 초기와 의존성, 오목한 경계(concavity) 분할의 비적합성, 그리고 모델내 요소간 자체교차(self-intersection)의 제한점을 가지고 있었다. 본 연구에서는 이러한 제한점을 극복하고, 오목한 구조를 포함하는 복잡한 인체기관의 경계를 분할하기에 적합한 새로운 변형모델을 제안하였다. 제안한 변형모델은 볼륨영상 피라미드(pyramid)를 기반으로 다해상도(multiresolution)의 모델 정제화(refinement)를 수행한다. 다해상도 모델 정제화는 전역적 시셈플링(global resampling) 및 지역적 리샘플링(local resampling)를 통하여 저해상도의 모델로부터 점차 고해상도의 모델로 이동하면서 객체의 경계를 계층적으로 분할해가는 방법이다. 다해상도 모델에 의한 계층적 경계 분할은 초기화 조건에의 의존성을 극복할 수 있게할 뿐 아니라, 빠른 속도로 원하는 객체의 경계에 수렴할 수 있게 한다. 또한 지역적 리샘플링은 모델 구성요소의 정규화를 수행함으로써 객체의 오목한 부분을 성공적으로 분할할 수 있게 한다. 그리고, 제안 모델은 기존 변형모델에서 포함하는 내부 힘(internal force)과 외부 힘(external force)외에 자체교차방지 힘(non-self-intersection force)을 추가함으로서 효과적으로 모델내의 자체교차를 방지할 수 있게 하였다.

• Journal of Korea Multimedia Society
• /
• v.22 no.3
• /
• pp.335-348
• /
• 2019

Cell segmentation and counting represent one of the most important tasks required in order to provide an exhaustive understanding of biological images. Conventional features suffer the lack of spatial consistency by causing the joining of the cells and, thus, complicating the cell counting task. We propose, in this work, a cascade of networks that take as inputs different versions of the original image. After constructing a Gaussian pyramid representation of the microscopy data, the inputs of different size and spatial resolution are given to a cascade of deep convolutional autoencoders whose task is to reconstruct the segmentation mask. The coarse masks obtained from the different networks are summed up in order to provide the final mask. The principal and main contribution of this work is to propose a novel method for the cell counting. Unlike the majority of the methods that use the obtained segmentation mask as the prior information for counting, we propose to utilize the hidden latent representations, often called the high-level features, as the inputs of a neural network based regressor. While the segmentation part of our method performs as good as the conventional deep learning methods, the proposed cell counting approach outperforms the state-of-the-art methods.

• Journal of Information Processing Systems
• /
• v.18 no.3
• /
• pp.411-427
• /
• 2022

Digging historical and cultural information from seals in ancient books is of great significance. However, ancient Chinese seal samples are scarce and carving methods are diverse, and traditional digital image processing methods based on greyscale have difficulty achieving superior segmentation and recognition performance. Recently, some deep learning algorithms have been proposed to address this problem; however, current neural networks are difficult to train owing to the lack of datasets. To solve the afore-mentioned problems, we proposed an SEL-RefineMask which combines selector of feature pyramid network (SEL-FPN) with RefineMask to segment and recognize seals. We designed an SEL-FPN to intelligently select a specific layer which represents different scales in the FPN and reduces the number of anchor frames. We performed experiments on some instance segmentation networks as the baseline method, and the top-1 segmentation result of 64.93% is 5.73% higher than that of humans. The top-1 result of the SEL-RefineMask network reached 67.96% which surpassed the baseline results. After segmentation, a vision transformer was used to recognize the segmentation output, and the accuracy reached 91%. Furthermore, a dataset of seals in ancient Chinese books (SACB) for segmentation and small seal font (SSF) for recognition were established which are publicly available on the website.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.121-126
• /
• 2004

An efficient multi face detection technique is proposed based on hierarchical context-free attention operators in which multiple faces are efficiently detected from a noisy and complex background. A noise-tolerant generalized symmetry transform (NTSGT) is applied hierarchically, as a context free attention operator, to the input pyramidal image for the high speed global location of the regions of face candidates (ROFCs) with a single mask. For the face verification, local NTGST is applied within each ROFC to confirm the existence of the detailed facial features. First, by globally applying NTGST which introduces the average pyramid method and focusing to the input image with complex background, ROFCs with recognizable resolution are detected robustly. Morphological operations are applied only to the each detected ROFCs to emphasize the facial features like eyes and lips. Then, eyes are detected by locally appling NTGST to the ROFCs and only faces are detected by verifying the existence of the geometrical features of the faces relatively to the location of eyes. The experimental results show that the proposed method can efficiently detect multiple faces from a noisy or complex background with 93.5% detection rate.

• Journal of the Institute of Convergence Signal Processing
• /
• v.2 no.4
• /
• pp.31-37
• /
• 2001

License plate recognition system for an exclusive bus-lane is made of 5 core parts which are vehicle detection, image acquisition individual character extraction, character recognition and data transmission. Among them, the accuracy of license plate extraction can bring effect significantly to the accuracy of a whole system recognition rate also the more exact extraction of license plate is required in various weather and environment conditions. Therefore in this paper we propose a plat extraction algorithm that makes pyramid structure to reduced the extraction processing time binarizes plate's template region using adaptive thresholding extracts candidate region containing plate, and verifies a final region using plate character distribution characteristics among the candidates. Experimenal results were exactly extracted the license plate region by using proposed method to the image obtained in an exclusive bus-lane with various weather and environment conditions.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.12C
• /
• pp.1231-1238
• /
• 2005

This paper presents a new fast encoding algorithm for image vector quantization that incorporates the partial distances of multiple features with a multidimensional look-up table (LUT). Although the methods which were proposed earlier use the multiple features, they handles the multiple features step by step in terms of searching order and calculating process. On the other hand, the proposed algorithm utilizes these features simultaneously with the LUT. This paper completely describes how to build the LUT with considering the boundary effect for feasible memory cost and how to terminate the current search by utilizing partial distances of the LUT Simulation results confirm the effectiveness of the proposed algorithm. When the codebook size is 256, the computational complexity of the proposed algorithm can be reduced by up to the $70\%$ of the operations required by the recently proposed alternatives such as the ordered Hadamard transform partial distance search (OHTPDS), the modified $L_2-norm$ pyramid ($M-L_2NP$), etc. With feasible preprocessing time and memory cost, the proposed algorithm reduces the computational complexity to below the $2.2\%$ of those required for the exhaustive full search (EFS) algorithm while preserving the same encoding quality as that of the EFS algorithm.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.3
• /
• pp.529-536
• /
• 2022

In the past two years, Severe Acute Respiratory Syndrome Coronavirus-2(SARS-CoV-2) has been hitting more and more to people. This paper proposes a novel U-Net Convolutional Neural Network to classify and segment COVID-19 lung CT images, which contains Sub Coding Block (SCB), Atrous Spatial Pyramid Pooling(ASPP) and Attention Gate(AG). Three different models such as FCN, U-Net and U-Net-SCB are designed to compare the proposed model and the best optimizer and atrous rate are chosen for the proposed model. The simulation results show that the proposed U-Net-MMFE has the best Dice segmentation coefficient of 94.79% for the COVID-19 CT scan digital image dataset compared with other segmentation models when atrous rate is 12 and the optimizer is Adam.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.195-206
• /
• 2022

Crack detection is essential for inspection of existing structures and crack segmentation based on deep learning is a significant solution. However, datasets are usually one of the key issues. When building a new dataset for deep learning, laborious and time-consuming annotation of a large number of crack images is an obstacle. The aim of this study is to develop an approach that can automatically select a small portion of the most informative crack images from a large pool in order to annotate them, not to label all crack images. An active learning method with difficulty learning mechanism for crack segmentation tasks is proposed. Experiments are carried out on a crack image dataset of a steel box girder, which contains 500 images of 320×320 size for training, 100 for validation, and 190 for testing. In active learning experiments, the 500 images for training are acted as unlabeled image. The acquisition function in our method is compared with traditional acquisition functions, i.e., Query-By-Committee (QBC), Entropy, and Core-set. Further, comparisons are made on four common segmentation networks: U-Net, DeepLabV3, Feature Pyramid Network (FPN), and PSPNet. The results show that when training occurs with 200 (40%) of the most informative crack images that are selected by our method, the four segmentation networks can achieve 92%-95% of the obtained performance when training takes place with 500 (100%) crack images. The acquisition function in our method shows more accurate measurements of informativeness for unlabeled crack images compared to the four traditional acquisition functions at most active learning stages. Our method can select the most informative images for annotation from many unlabeled crack images automatically and accurately. Additionally, the dataset built after selecting 40% of all crack images can support crack segmentation networks that perform more than 92% when all the images are used.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.24 no.4
• /
• pp.372-378
• /
• 2014

Noise in biomedical X-ray image degrades the quality so that it might causes to decrease the accuracy of diagnosis. Especially the noise reduction techniques is quite essential for low-dose biomedical X-ray images obtained from low radiation power in order to protect patients, because their noise level is usually high to well discriminate objects. This paper proposes an efficient method to remove the noise in low-dose X-ray images while preserving the edges with diverse resolutions. In the proposed method, a noisy image is at first decomposed into several images with different resolutions in pyramidal representation, then the stable map of edge confidence is obtained from each of analyzed image using a fuzzy logic-based edge detector. This map is used to adaptively determine the parameter for guided filters, which eliminate the noise while preserving edges in the corresponding image. The filtered images in the pyramid are extended and synthesized into a resulted image using interpolation technique. The superiority of proposed method compared to the median, bilateral, and guided filters has been experimentally shown in terms of noise removal and edge preserving properties.