• Journal of IKEEE
• /
• 제27권3호
• /
• pp.273-279
• /
• 2023

The use of line memory is essential for image filtering in image processing hardware. After input data is stored in line memory, filtering is performed after synchronization to use the stored data. A sync generator is used for synchronization, and in the case of a conventional sync generator, the input sync signal is delayed by one row of the input image. If a signal delayed by two rows is required, it is necessary to connect two modules. This approach increases the size of the hardware and cannot be designed efficiently. In this paper, we propose a sync generator that generates multiple types of delayed signals by adding a finite state machine. The hardware design was coded in Verilog HDL, and performance is verified by applying it to image processing hardware using field programmable gate array board.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• 제8권7호
• /
• pp.1393-1398
• /
• 2004

Image is degraded by several causes such as the process of acquisition, storage and transmission. To restore those images, many researches have been continued. Centrally methods to restore degraded image by AWGN(additive white gaussian noise) a.e mean filter and wiener filter. Especially, mean filter is superior in noise reduction of area that is a small change of luminosity. But mean filter brings about the effect smoothing edge components of the image, because it does'nt consider characteristics of the image. So in this paper we propose an image restoration method compounding respective images adding established weights, after filtering with mean filter and powerful wiener filter in both improvement of contrast and preservation of edge components.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• 제6권5호
• /
• pp.789-795
• /
• 2002

Color image is formed of combination of three color channels. Therefore its architecture is very complicated and it requires complicated image Processing for effective image segmentation. In this paper. we propose architecture of universalized Color Morphological Pyramids(CMP) which is able to give effective image segmentation. Image Pyramid architecture is a successive Image sequence whose area ratio $2^{\int}({\int}=1,2,....,N)$ after filtering and subsampling of input image. In this technique, noise removed by sequential filtering and resolution is degraded by downsampling using CMP in various color spaces. After that, new level images are constructed that apply formula using distance of neighbor vectors in close level images and segments its image. The feasibility of proposed method is examined by comparing with the results obtained from the existing method.

• The KIPS Transactions:PartB
• /
• 제15B권2호
• /
• pp.123-130
• /
• 2008

Automatic lipreading is to recognize speech by observing the movement of a speaker's lips. It has received attention recently as a method of complementing performance degradation of acoustic speech recognition in acoustically noisy environments. One of the important issues in automatic lipreading is to define and extract salient features from the recorded images. In this paper, we propose a feature extraction method by using a new filtering technique for obtaining improved recognition performance. The proposed method eliminates frequency components which are too slow or too fast compared to the relevant speech information by applying a band-pass filter to the temporal trajectory of each pixel in the images containing the lip region and, then, features are extracted by principal component analysis. We show that the proposed method produces improved performance in both clean and visually noisy conditions via speaker-independent recognition experiments.

• Journal of the Institute of Convergence Signal Processing
• /
• 제11권2호
• /
• pp.151-156
• /
• 2010

This paper is a study on image noise reduction of modified sigma filter by image decomposition using directivity. Conventional sigma filter has been shown to be a good solution both in terms of filtering accuracy and computational complexity. However, the sigma filter does not preserve well small edges especially for high level of additive noise. In this paper, we propose here a new method using a modified sigma filter. In our proposed method the input image is first decomposed in two components that have features of horizontal, vertical and diagonal direction. Then, two components are applied HPF and LPF. By applying a conventional sigma filter separately on each of them, the output image is reconstructed from the filtered components. Added noise is removed and our proposed method preserves the edges from the image. Comparative results from experiments show that the proposed algorithm achieves higher gains, on average, 2.6 dB PSNR than the sigma filter and 0.5 dB PSNR than the modified sigma filter. When relatively high levels of noise added, the proposed algorithm shows better performance than two conventional filters.

• 한국HCI학회:학술대회논문집
• /
• 한국HCI학회 2006년도 학술대회 1부
• /
• pp.1289-1297
• /
• 2006

모든 곳에 존재하는 네트워크 환경을 의미하는 '유비쿼터스' 시대와 최신 기술로 구현되어 인간을 도와주는 '지능형 로봇'의 시대가 도래하고 있다. 기술의 흐름은, 이제 우리에게 공장과 공원 등의 공공 장소뿐 만이 아니라, 생활의 기본이 되는 가정 안에서의 로봇을 받아들일 준비를 요구하고 있다. 로봇과 사용자는 실제 생활 속에서 많은 상호 작용을 하게 되며, 필연적으로 여러 가지의 불확실성을 내포하게 되는데, 각각의 요청들과 상황들은, 미리 정해진 규칙에 의거해 처리하기에는 너무 다양하다. 그 어려움을 극복하는 방법으로, 어떤 상황에 적응하는 방법으로 기억을 사용 하는 인간과 마찬가지로, 로봇은 새로운 요청을 처리하기 위해 과거의 기록을 사용할 수 있다. 여러 가지 과거의 기록들을 잘 정리해서 분류하여 저장해둔 후, 현재의 요청에 대한 답으로, 가장 가능성 있는 과거의 기록을 찾아내는 것이다. 본 논문에서는 사용자와 로봇 사이에서 상호 작용에서 발생할 수 있는 불확실성을 과거기록의 탐색을 통해 해결하고자 하였다. 과거 기록은 시간, 장소, 대상 물건, 행동 유형으로 구분되어 저장하였으며, 각각의 유사 가능성(Possibility)들의 합을 기준으로, 전체 기록을 K-Means 알고리즘을 통하여 군집화하고 협력 필터링을 기반으로 현재의 요청이 담고 있는 불확실성에 대한 가능성 있는 값을 추천해 주었다. 제한된 공간과 제한된 자료의 수에 의한 실험 결과로서의 한계를 가지고 있지만, 실제 가정용 로봇에서의 적용 가능성을 보여주었다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• 제14권11호
• /
• pp.2413-2418
• /
• 2010

We propose a retouching method that converts a photography taken by smartphone to a watercolor painting image using bilateral filtering and mean shift segmentation which are mostly used in image processing. The first step is to convert an input image to fit the screen resolution of smartphone. And next step is to weaken high frequency components of the image, while preserving the edge of image using the bilateral filtering. And after that we perform mean shift segmentation from the bilateral filtered image. We apply parameters of mean shift segmentation considering the processing speed of smartphone. Experimental result shows that our method can be applied to various types of image and bring better result.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 한국해양정보통신학회 2010년도 추계학술대회
• /
• pp.206-208
• /
• 2010

We propose a retouching method that converts a photography taken by smartphone to a watercolor painting image using bilateral filtering and mean shift segmentation which are mostly used in image processing. The first step is to convert an input image to fit the screen resolution of smartphone. And next step is to weaken high frequency components of the image, while preserving the edge of image using the bilateral filtering. And after that we perform mean shift segmentation from the bilateral filtered image. We apply parameters of mean shift segmentation considering the processing speed of smartphone. Experimental result shows that our method can be applied to various types of image and bring better result.

• Journal of the Korea Society of Computer and Information
• /
• 제15권9호
• /
• pp.25-33
• /
• 2010

We propose a retouching method that converts a general photography to a watercolor painting image using bilateral filtering and mean shift segmentation which are mostly used in image processing. The first step is to weaken high frequency components of the image, while preserving the edge of image using the bilateral filtering. And after that we perform DoG(Difference of Gradient) edge extraction and mean shift segmentation respectively from the bilateral filtered image. The DoG edge extraction is performed using luminance component of the image whose RGB color space is transformed into CIELAB space. Experimental result shows that our method can be applied to various types of image and bring better result, especially against the photo taken in daylight.

• Journal of the Korea Convergence Society
• /
• 제12권5호
• /
• pp.1-7
• /
• 2021

In the latest trend of machine translation research, the model is pretrained through a large mono lingual corpus and then finetuned with a parallel corpus. Although many studies tend to increase the amount of data used in the pretraining stage, it is hard to say that the amount of data must be increased to improve machine translation performance. In this study, through an experiment based on the mBART model using parallel corpus filtering, we propose that high quality data can yield better machine translation performance, even utilizing smaller amount of data. We propose that it is important to consider the quality of data rather than the amount of data, and it can be used as a guideline for building a training corpus.