• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.11a
• /
• pp.1-3
• /
• 2022

최근 기존의 영상 압축 파이프라인 대신 신경망의 종단 간 학습을 통해 압축을 수행하는 알고리즘의 연구가 활발히 진행되고 있다. 본 논문은 종단 간 학습 기반 공간적 스케일러블 압축 기술을 제안한다. 보다 구체적으로 본 논문은 신경망의 각 계층에서 하위 계층의 학습된 특징 (feature)을 융합하여 상위 계층으로 전달하는 다중 스케일 특징 융합 (multi-scale feature fusion) 모듈을 도입해 상위 계층이 더욱 풍부한 특징 정보를 학습하고 계층 사이의 특징 중복성을 더욱 잘 제거할 수 있도록 한다. 기존 방법 대비 향상 계층(enhancement layer)에서 1.37%의 BD-rate가 향상된 결과를 볼 수 있다.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.5 no.2
• /
• pp.16-24
• /
• 2002

Classification of the land cover characteristics is a major application of remote sensing. The goal of this study is to propose an optimal classification process for electro-optical camera(EOC) of Korea Multi-Purpose Satellite(KOMPSAT). The study was carried out on Landsat TM, high spectral resolution image and KOMPSAT EOC, high spatial resolution image of Miho river basin, Korea. The study was conducted in two stages: one was image fusion of TM and EOC to gain high spectral and spatial resolution image, the other was land cover classification on fused image. Four fusion techniques were applied and compared for its topographic interpretation such as IHS, HPF, CN and wavelet transform. The fused images were classified by radial basis function neural network(RBF-NN) and artificial neural network(ANN) classification model. The proposed RBF-NN was validated for the study area and the optimal model structure and parameter were respectively identified for different input band combinations. The results of the study propose an optimal classification process of KOMPSAT EOC to improve the thematic mapping and extraction of environmental information.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.8
• /
• pp.2068-2082
• /
• 2023

With the rapid development of deep learning, Depth Map Super-Resolution (DMSR) method has achieved more advanced performances. However, when the upsampling rate is very large, it is difficult to capture the structural consistency between color features and depth features by these DMSR methods. Therefore, we propose a color-image guided DMSR method based on iterative depth feature enhancement. Considering the feature difference between high-quality color features and low-quality depth features, we propose to decompose the depth features into High-Frequency (HF) and Low-Frequency (LF) components. Due to structural homogeneity of depth HF components and HF color features, only HF color features are used to enhance the depth HF features without using the LF color features. Before the HF and LF depth feature decomposition, the LF component of the previous depth decomposition and the updated HF component are combined together. After decomposing and reorganizing recursively-updated features, we combine all the depth LF features with the final updated depth HF features to obtain the enhanced-depth features. Next, the enhanced-depth features are input into the multistage depth map fusion reconstruction block, in which the cross enhancement module is introduced into the reconstruction block to fully mine the spatial correlation of depth map by interleaving various features between different convolution groups. Experimental results can show that the two objective assessments of root mean square error and mean absolute deviation of the proposed method are superior to those of many latest DMSR methods.

• Journal of Broadcast Engineering
• /
• v.22 no.6
• /
• pp.702-712
• /
• 2017

In this paper, we propose an unsharp masking algorithm using Laplacian as a weight map for the signal structure and a gamma transformation algorithm using image mean intensity as a weight map for mean intensity. The conventional weight map based on the patch has a disadvantage in that the brightness in the image is shifted to one side in the signal structure and the mean intensity region. So the detailed information is lost. In this paper, we improved the detail using unsharp masking of patch unit and proposed linearly combined the gamma transformed values using the average brightness values of the global and local images. Through the proposed algorithm, the detail information such as edges are preserved and the subjective image quality is improved by adjusting the brightness of the light. Experiment results show that the proposed algorithm show better performance than conventional algorithm.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.189-190
• /
• 2016

일반 영상의 영상확대를 위한 다양한 알고리즘이 존재한다. 하지만 적외선 열화상 영상의 경우 일반영상과 다른 특성을 가지고 있기 때문에 적외선 영상을 위한 영상 확대 알고리즘이 필요하다. 따라서 적외선 영상이 일반영상에 비해 디테일이 없다는 특성을 고려하여 복잡한 알고리즘을 적용시키기 보다는 ADRC 와 같은 단순한 분류 기법을 활용하여 LR-HR 패치를 분류하고 학습된 데이터를 이용하여 영상확대 알고리즘에 적용하였다. 알고리즘의 성능 향상을 위해 학습과정에 전처리 과정을 추가하여 합성과정에서 추가적인 연산량의 증가 없이 확대 영상의 선명도를 향상시키고자 하였다. 또한 확대된 적외선 영상이 동일 해상도의 가시광영상에 비해 선명도가 떨어진다는 점을 고려하여 확대된 적외선 영상에 가시광영상의 고주파 정보를 합성시켜 이전보다 영상의 선명도를 더 향상시키고자 하였다. 이와 같은 방법으로 영상 확대 알고리즘만 수행하였을 때 통상적인 영상확대 기법인 bi-cubic interpolation 기법보다 JNB 수치가 평균 0.0727 만큼 높은 결과를 확인할 수 있었고 가시광영상과 융합하였을 때 이전보다 평균 0.0742 만큼 더 선명해진 영상을 얻었다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.8 no.3
• /
• pp.95-103
• /
• 1998

This paper deals with a problem occuring in recognition of tactile images due to the effects of imposed force at a me urement moment. Tactile image of a contact surface, used for recognition of the surface type, varies depending on the forces imposed so that a false recognition may result in. This paper fuzzifies two parameters of the contour of a tactile image with the membership function formed by considering the imposed force. Two fuzzifed paramenters are fused by the average Minkowski's dist; lnce. The proposed algorithm was implemented on the multisensor system cnmposed of an optical tact le sensor and a 6 axes forceltorque sensor. By the experiments, the proposed algorithm has shown average recognition ratio greater than 869% over all imposed force ranges and object models which is about 14% enhancement comparing to the case where only the contour information is used. The pro- ~oseda lgorithm can be used for end-effectors manipulating a deformable or fragile objects or for recognition of 3D objects by implementing on multi-fingered robot hand.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37C no.9
• /
• pp.751-756
• /
• 2012

In this paper, we propose a foreground extraction and depth map generation method using a time-of-flight (TOF) depth camera. Although, the TOF depth camera captures the scene's depth information in real-time, it has a built-in noise and distortion. Therefore, we perform several preprocessing steps such as image enhancement, segmentation, and 3D warping, and then use the TOF depth data to generate the depth-discontinuity regions. Then, we extract the foreground object and generate the depth map as of the color image. The experimental results show that the proposed method efficiently generates the depth map even for the object boundary and textureless regions.

• Journal of the Korean Society of Environmental Restoration Technology
• /
• v.9 no.6
• /
• pp.1-12
• /
• 2006

Remote sensing cannot provide a direct measurement of vegetation index (VI) but it can provide a reasonably good estimate of vegetation index, defined as the ratio of satellite bands. The monitoring of vegetation in nearby urban regions is made difficult by the low spatial resolution and temporal resolution image captures. In this study, enhancing spatial resolution method is adapted as to improve a low spatial resolution. Recent studies have successfully estimated normalized difference vegetation index (NDVI) using improved resolution method such as from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard EOS Terra satellite. Image enhancing spatial resolution is an important tool in remote sensing, as many Earth observation satellites provide both high-resolution and low-resolution multi-spectral images. Examples of enhancement of a MODIS multi-spectral image and a MODIS NDVI image of Cheongju using a Landsat TM high-resolution multi-spectral image are presented. The results are compared with that of the IHS technique is presented for enhancing spatial resolution of multi-spectral bands using a higher resolution data set. To provide a continuous monitoring capability for NDVI, in situ measurements of NDVI from paddy field was carried out in 2004 for comparison with remotely sensed MODIS data. We compare and discuss NDVI estimates from MODIS sensors and in-situ spectroradiometer data over Ochang plain region. These results indicate that the MODIS NDVI is underestimated by approximately 50%.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.1 no.1
• /
• pp.109-117
• /
• 1998

To obtain new information using a single remotely sensed image data is limited to extract various information. Recent trends in the remote sensing show that many researchers integrate and analyze many different forms of remotely sensed data, such as optical and radar satellite images, aerial photograph, airborne multispectral scanner data and land spectral scanners. Korean researchers have not been using such a combined dataset yet. This study intended to apply the technique of integration between optical data and radar data(SAR) and to examine the output that had been obtained through the technique of supervised classification using the result of integration. As a result, we found of better enhanced image classification results by using IHS conversion than by using RGB mixed and interband correlation.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.45 no.5
• /
• pp.162-172
• /
• 2008

In this paper, we propose a multimodal speaker verification method using teeth image and voice as biometric trait for personal verification in mobile terminal equipment. The proposed method obtains the biometric traits using image and sound input devices of smart-phone that is one of mobile terminal equipments, and performs verification with biometric traits. In addition, the proposed method consists the multimodal-fashion of combining two biometric authentication scores for totally performance enhancement, the fusion method is accompanied a weighted-summation method which has comparative simple structure and superior performance for considering limited resources of system. The performance evaluation of proposed multimodal speaker authentication system conducts using a database acquired in smart-phone for 40 subjects. The experimental result shows 8.59% of EER in case of teeth verification 11.73% in case of voice verification and the multimodal speaker authentication result presented the 4.05% of EER. In the experimental result, we obtain the enhanced performance more than each using teeth and voice by using the simple weight-summation method in the multimodal speaker verification system.