• Journal of the Institute of Convergence Signal Processing
• /
• v.13 no.2
• /
• pp.63-69
• /
• 2012

Noise is regarded as an unwanted component of the image because it significantly reduces image quality. And image is often corrupted by mixed noise. In this paper an efficient modified weighted filter algorithm which combines spatial weight and intensity weight is proposed for removing mixed noise. In the proposed method, the filtering mask is separated into the four sub-windows and the parameters of the weights are confirmed by calculating local standard deviation and the mean of four sub-windows' standard deviations. Considering the spatial information and intensity information, the proposed method has good performance on not only noise elimination but also preservation of details. Simulation results demonstrate that the proposed method performs better than conventional algorithms.

• Journal of the Korean Society of Industry Convergence
• /
• v.21 no.6
• /
• pp.389-394
• /
• 2018

Remotely operated vehicle(ROV) and autonomous underwater vehicle(AUV) have been used for underwater surveys, underwater exploration, resource harvesting, offshore plant maintenance and repair, and underwater construction. It is hard for people to work in the deep sea. Therefore, we need a vision control system of underwater submersible that can replace human eyes. However, many people have difficulty in developing a deep-sea image control system due to the deep sea special environment such as high pressure, brine, waterproofing and communication. In this paper, we will develop an Ethernet based remote image control system that can control the image mounted on ROV.

• International journal of advanced smart convergence
• /
• v.8 no.1
• /
• pp.18-23
• /
• 2019

The image projected onto the screen of the floating hologram is no more than a two-dimensional image. Although it creates an illusion that an object appears to float in space as it moves around while showing its different parts. This paper has proposed a novel method of floating 3D hologram display to view stereoscopic three-dimensional images without putting on glasses. The system is comprised of a sharkstooth scrim screen, projector, polarizing filter for the projector, and a polarizing film to block the image projected from the sham screen. As part of the polarization characteristics, the background image and the front object have completely been separated from each other with the stereoscopic 3D effect successfully implemented by the binocular disparity caused by the distance between the two screens.

• Journal of information and communication convergence engineering
• /
• v.21 no.1
• /
• pp.1-8
• /
• 2023

This paper presents a design method for a wideband bandpass filter (BPF) which compensates for frequency dependency based on the image admittance and image phase. In the proposed method, new compensation methods for the admittance and phase are integrated with the conventional method. The proposed method improves the frequency shift and reduces the unwanted bandwidth when designing more than 20% of the Fractional Bandwidth (FBW), whereas the conventional method exhibits frequency degradation at only 10% FBW. The proposed design theory was verified by applying it to both lumped elements and distributed lines through circuit simulation and measurements without an optimization process. The measurement results demonstrate improvements in the frequency shift and target bandwidth. In the future, an accurate design method based on frequency dependence can be implemented for the next-generation broadband communication system applications.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.332-333
• /
• 2021

In this paper, we propose a new image classification method based on several trainings, which is mainly used to solve model overfitting and non-convergence in image classification tasks of small data sets and to improve classification accuracy.

• Smart Media Journal
• /
• v.12 no.9
• /
• pp.45-59
• /
• 2023

Weakly supervised object localization (WSOL) is a task of localizing an object in an image using only image-level labels. Previous studies have followed the conventional class activation mapping (CAM) pipeline. However, we reveal the current CAM approach suffers from problems which cause original CAM could not capture the complete defects features. This work utilizes a convolutional neural network (CNN) pretrained on image-level labels to generate class activation maps in a multi-scale manner to highlight discriminative regions. Additionally, a vision transformer (ViT) pretrained was treated to produce multi-head attention maps as an auxiliary detector. By integrating the CNN-based CAMs and attention maps, our approach localizes defective regions without requiring bounding box or pixel-level supervision during training. We evaluate our approach on a dataset of apple images with only image-level labels of defect categories. Experiments demonstrate our proposed method aligns with several Object Detection models performance, hold a promise for improving localization.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.225-228
• /
• 2001

We propose a method which has effective scalable indexing characteristics of large color image archives. Color key and energy key method which uses provides high performance of fast accessing and has almost constant query response time over DH size.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.12a
• /
• pp.129-132
• /
• 2000

In this paper, we deal with content based searching algorithm for brand image using GHA(Generalized Hebbian Algorithm). GHA is a part of PCA(Principal Component Analysis), that has single-layer perceptron operates and self-organizing performances. We used this algorithm for feature extracts of brand images, and our simulations verify the high performance than present text based methods.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.08a
• /
• pp.201-204
• /
• 2000

본 논문에서는 기존의 칼라 항상성(color constancy) 알고리즘을 기반으로 한 새로운 영상 복원 방법을 제안한다. 이 방법은 인간의 시각 특성을 선형 모델로 표현한 칼라 항상성 모델을 이용한다. 칼라 항상성 모델에서 가장 중요한 과정은 영상의 광원을 추정하는 것이다. 이론 위하여 영상에서 명도 값이 가장 큰 화소의 값을 이용하며, 이 값을 각 수광체(photoreceptor)의 응답으로 대체한다. 추정된 광원을 이용하여 만들어진 영상을 일정한 비율로 스케일링하여 영상을 복원한다. 실험을 통하여 제안한 항법으로 영상 복원이 잘 됨을 알 수 있었다.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.08a
• /
• pp.205-208
• /
• 2000

문자영상의 인식분야에서는 다양한 잡음으로 인해 손상된 영상을 개선하는 영상의 전처리 과정을 아주 중요시하고 있다. 본 논문에서는 영상의 객체 윤곽을 보호하면서 충격잡음에 효과적인 비선형 중간값(median) 필터를 먼저 적용하고, 영상객체에 대해 선택적으로 축소하고 확대하는 형태학(morphological) 필터를 직렬로 조합함으로서 선명한 문자 영상을 획득하였고, 좀 더 현실적인 문자를 인식하도록 Genesis 영상처리 보드와 이를 제어하는 MIL(Matrox Image Library) 패키지를 사용하여 실시간적으로 구현해 보았다.