• The Journal of the Acoustical Society of Korea
• /
• v.15 no.1
• /
• pp.54-60
• /
• 1996

In most pattern recognition and image understanding applications, images are degraded by noise and other distortions. Therefore, it is more relevant to decide how similar two objects are rather than to decide whether the two are exactly the same. In this paper, we propose a method for recognizing degraded symbols using a distance measure between two graphs representing the symbols. a symbol is represented as a graph consisting of nodes and edges based on the run graph concept. The graph is then transformed into a reference model graph with production rule containing the embedding transform. The symbols are recognized by using the distance measure which is estimated by using the number of production rules used and the structural homomorphism between a transformed graph and a model graph. the proposed approach is applies to the recognition of non-note musical symbols and the result are given.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.2
• /
• pp.216-222
• /
• 2008

This paper proposed the multi characters detection algorithm using Color segmentation and the closing curve feature of LoG Operator in order to complement the demerit of the existing research which is weak in complexity of background, variety of light and disordered line and similarity of left and background color, etc. The proposed multi characters detection algorithm divided into three parts : The feature detection, characters format and characters detection Parts in order to be possible to apply to image of various feature. After preprocess that the new multi characters detection algorithm that proposed in this paper used wavelet, morphology, hough transform which is the synthesis logical model in order to raise detection rate by acquiring the non-perfection characters as well as the perfection characters with processing OR operation after processing each color area by AND operation sequentially. And the proposal algorithm is simulated with natural images which include natural character area regardless of size, resolution and slant and so on of image. And the proposal algorithm in this paper is confirmed to an excellent detection rate by compared with the conventional detection algorithm in same image.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.47 no.3
• /
• pp.1-10
• /
• 2010

As various interfacing devices for computational machines are being developed, a new HCI method using hand motion input is introduced. This interface method is a vision-based approach using a single camera for detecting and tracking hand movements. In the previous researches, only a skin color is used for detecting and tracking hand location. However, in our design, skin color and shape information are collectively considered. Consequently, detection ability of a hand increased. we proposed primary orientation edge descriptor for getting an edge information. This method uses only one hand model. Therefore, we do not need training processing time. This system consists of a detecting part and a tracking part for efficient processing. In tracking part, the system is quite robust on the orientation of the hand. The system is applied to recognize a hand written number in script style using DNAC algorithm. Performance of the proposed algorithm reaches 82% recognition ratio in detecting hand region and 90% in recognizing a written number in script style.

• The KIPS Transactions:PartD
• /
• v.15D no.4
• /
• pp.581-588
• /
• 2008

Old films or photographs usually have damages from physical or chemical effects, and the damage and digitalization make stain, scratch, scribbling, noise, and digital drop out in frames. Damages include global damage and local damage, and it is well known that local damage restoration is a main factor for improving image quality. Previous researches have focused on impairment localization (esp. for line scratch impairments) and restoration techniques for line scratch, dirt, blob, and intentional scratch. Inpainting is a key technique using partial derivatives to restore damages in images. It does not show good quality for the complex images because it is based on finite order for partial derivatives, and it takes much time complexity. In this paper, we present a modified inpainting scheme, where we use Sobel edge operator's and angle to compute isophotes, and compare our scheme with Bertalmio's scheme. We experiment our scheme with two old Korean films, and Simulation results show that our scheme requires smaller time complexity than Bertalmio's scheme with comparable reconstructed image quality.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.5
• /
• pp.347-352
• /
• 2013

In impulse noise filtering techniques window size play an important role. Usually, an appropriate window is determined according to the noise density. A small window may not be able to suppress noise properly whereas a large window may remove edges and fine image details. Moreover, the value of the central pixel is estimated by considering all pixels within the window. In this work, contrary to the previous approaches, we propose an iterative impulse noise removal scheme that emphasizes on noise-free pixels within a small neighborhood. The iterative process continues until all noisy pixels are replaced with the estimated pixels. In order to estimate the optimal value for a noisy pixel, a genetic programming (GP) based estimator is evolved that takes few noise-free pixels as input. The estimator is constituent of noise-free pixels, arithmetic operators and random constants. Experimental results show that theproposed scheme is capable of removing impulse noise effectively while preserving the fine image details. Especially, our approach has shown effectiveness against high impulse noise density.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.11
• /
• pp.549-554
• /
• 2016

ConceptNet is a common sense knowledge base which is formed in a semantic graph whose nodes represent concepts and edges show relationships between concepts. As it is difficult to make knowledge base integrity, a knowledge base often suffers from incompleteness problem. Therefore the quality of reasoning performed over such knowledge bases is sometimes unreliable. This work presents neural tensor networks which can alleviate the problem of knowledge bases incompleteness by reasoning new assertions and adding them into ConceptNet. The neural tensor networks are trained with a collection of assertions extracted from ConceptNet. The input of the networks is two concepts, and the output is the confidence score, telling how possible the connection between two concepts is under a specified relationship. The neural tensor networks can expand the usefulness of ConceptNet by increasing the degree of nodes. The accuracy of the neural tensor networks is 87.7% on testing data set. Also the neural tensor networks can predict a new assertion which does not exist in ConceptNet with an accuracy 85.01%.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.6
• /
• pp.313-330
• /
• 2021

In an autonomous driving system equipped with a camera and lidar, depth completion techniques enable dense depth estimation. In particular, using self-supervised learning it is possible to train the depth completion network even without ground truth. In actual autonomous driving, such depth completion should have very short latency as it is the input of other algorithms. So, rather than complicate the network structure to increase the accuracy like previous studies, this paper focuses on network latency. We design a U-Net type network with RegNet encoders optimized for GPU computation. Instead, this paper presents several techniques that can increase accuracy during the process of self-supervised learning. The proposed techniques increase the robustness to unreliable lidar inputs. Also, they improve the depth quality for edge and sky regions based on the semantic information extracted in advance. Our experiments confirm that our model is very lightweight (2.42 ms at 1280x480) but resistant to noise and has qualities close to the latest studies.

• Journal of Digital Policy
• /
• v.1 no.2
• /
• pp.53-59
• /
• 2022

In this paper, we proposed a method using sub-edge information extracted through a hierarchical structure in the process of generating super resolution based on a single image. In order to improve the quality of super resolution, it is necessary to clearly distinguish the shape of each area while clearly expressing the boundary area in the image. The proposed method assists edge information of the image in deep learning based super resolution method to create an improved super resolution result while maintaining the structural shape of the boundary region, which is an important factor determining the quality in the super resolution process. In addition to the group convolution structure for performing deep learning based super resolution, a separate hierarchical edge accumulation extraction process based on high-frequency band information for sub-edge extraction is proposed, and a method of using it as an auxiliary feature is proposed. Experimental results showed about 1% performance improvement in PSNR and SSIM compared to the existing super resolution.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.2C
• /
• pp.141-149
• /
• 2007

Associating the shoulder line with head location of the human body is useful in verifying, localizing and tracking persons in an image. Since the head line and the shoulder line, what we call ${\Omega}$-shape, move together in a consistent way within a limited range of deformation, we can build a statistical shape model using Active Shape Model (ASM). However, when the conventional ASM is applied to ${\Omega}$-shape fitting, it is very sensitive to background edges and clutter because it relies only on the local edge or gradient. Even though appearance is a good alternative feature for matching the target object to image, it is difficult to learn the appearance of the ${\Omega}$-shape because of the significant difference between people's skin, hair and clothes, and because appearance does not remain the same throughout the entire video. Therefore, instead of teaming appearance or updating appearance as it changes, we model the discriminative appearance where each pixel is classified into head, torso and background classes, and update the classifier to obtain the appropriate discriminative appearance in the current frame. Accordingly, we make use of two features in fitting ${\Omega}$-shape, edge gradient which is used for localization, and discriminative appearance which contributes to stability of the tracker. The simulation results show that the proposed method is very robust to pose change, occlusion, and illumination change in tracking the head and shoulder line of people. Another advantage is that the proposed method operates in real time.

• Journal of Broadcast Engineering
• /
• v.17 no.5
• /
• pp.756-764
• /
• 2012

In current HEVC, prediction pixels are finally calculated by linear-function interpolation on two reference pixels. It is hard to expect good performance on the case of occurring large difference between two reference pixels. This paper decides more accurate prediction pixel values than current HEVC using linear function. While existing prediction process only uses two reference pixels, proposed method uses DCT-IF. DCT-IF analyses frequency characteristics of more than two reference pixels in frequency domain. And proposed method calculates prediction value adaptively by using linear-function, DCT-IF and cubic-function to decide more accurate interpolation value than to only use linear function. Cubic-function has a steep slope than linear-function. So, using cubic-function is utilized on edge in prediction unit. The complexity of encoder and decoder in HM6.0 has increased 3% and 1%, respectively. BD-rate has decreased 0.4% in luma signal Y, 0.3% in chroma signal U and 0.3% in chroma signal V in average. Through this experiment, proposed adaptive intra prediction method using DCT-IF and cubic-function shows increased performance than HM6.0.