• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.47 no.1
• /
• pp.160-170
• /
• 2010

This paper proposes a Haar Wavelet Feature Detector (HWFD) based on the Haar wavelet transform and average box filter. By decomposing the original image using the Haar wavelet transform, the proposed detector obtains the variance information of the image, making it possible to extract more distinctive features from the original image. For detection of interest points that represent the regions whose variance is the highest among their neighbor regions, we apply the average box filter to evaluate the local variance information and use the integral image technique for fast computation. Due to utilization of the Haar wavelet transform and the average box filter, the proposed detector is robust to illumination change, scale change, and rotation of the image. Experimental results show that even though the proposed method detects fewer interest points, it achieves higher repeatability, higher efficiency and higher matching accuracy compared with the DoG detector and Harris corner detector.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.6
• /
• pp.73-80
• /
• 1996

Success of a face recognition system depends on which parameters are used. Generally the parameters are affected by environment of facial images such as illumination. To reduce the influence of the evcironment, since side images are insensitive to variance of brightness, it might be an appropriate approach to make the defect of front face images complete with the features extracted from side images. This paper proposes a method which collects and completes the information of front and side images. It is intended to prove the usefulness of the method that it is compared with other methods.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10b
• /
• pp.98-101
• /
• 2006

본 논문에서는 얼굴영상 인식의 전처리 단계인 피부 영역 자동 검출시 적용 가능하며 조명변화에 강건한 피부 영역 검출을 위한 혼합 컬러모델을 제시한다. 또한, 사용자별로 차이를 보이는 다양한 피부색을 자동으로 인식하고 사용자에 적합한 피부색 영역을 결정하기 위하여 제시한 컬러 모델을 기반으로 한 피부색 영역 모델링 전처리 단계를 제시한다. 우선, 사용자 및 사용 카메라에 따라 차이를 보이는 피부색에 대한 영역 모델을 구축하기 위하여 화면상의 가운데에 손이나 얼굴 영역이 위치하도록 하고 일정 프레임의 화면 정보를 취득한다. 취득 화면 정보로서 각 픽셀에 대한 정규화 된 RGB 성분 및 H 성분, V 성분 정보를 추출하고 이에 대한 평균화된 혼합 컬러 모델을 구축한다. H성분으로 피부색과 비슷한 배경을 제거하고 여기에 YUV 성분 중 적색에서 밝기 값을 뺀 성분인 V 값을 한 번 더 사용하여 밝기 값을 제거한 보다 뚜렷한 얼굴영역을 검출한다.

• The Journal of Korea Robotics Society
• /
• v.17 no.3
• /
• pp.314-321
• /
• 2022

Depth information used in SLAM and visual odometry is essential in robotics. Depth information often obtained from sensors or learned by networks. While learning-based methods have gained popularity, they are mostly limited to RGB images. However, the limitation of RGB images occurs in visually derailed environments. Thermal cameras are in the spotlight as a way to solve these problems. Unlike RGB images, thermal images reliably perceive the environment regardless of the illumination variance but show lacking contrast and texture. This low contrast in the thermal image prohibits an algorithm from effectively learning the underlying scene details. To tackle these challenges, we propose multi-channel remapping for contrast. Our method allows a learning-based depth prediction model to have an accurate depth prediction even in low light conditions. We validate the feasibility and show that our multi-channel remapping method outperforms the existing methods both visually and quantitatively over our dataset.

• Journal of the Korea Computer Graphics Society
• /
• v.16 no.3
• /
• pp.41-48
• /
• 2010

Photon mapping is a traditional global illumination method using many photons emitted from the light source for photo-realistic rendering. However, this method needs a lot of resources to perform tracing of millions of photons. Progressive photon mapping solves this problem. Typical progressive photon mapping performs ray tracing at first to find the hit points on diffuse surface of objects. Next, light source repeatedly emits a small number of photons in photon tracing pass, and power of photons in each sphere that has a fixed radius with the hit points in the center is accumulated. This method requires less resources than previous photon mapping, but it spends much time for gathering enough photons since each of photons progresses through a random direction and rendering high quality image. To improve the method, we propose photon probing that calculates variance of photons in the sphere and controls radius of sphere. In addition, we apply cone filter in radiance estimation step for reducing aliasing at the edges in result image.

• Journal of KIISE:Software and Applications
• /
• v.29 no.11
• /
• pp.809-817
• /
• 2002

This paper describes a new face detection method that is a pre-processing algorithm for on-line face recognition. To complement the weakness of using only edge or rotor features from previous face detection method, we propose the two types of face detection method. The one is a combined method with edge and color features and the other is a center area color sampling method. To prevent connecting the people's face area and the background area, which have same colors, we propose a new adaptive edge detection algorithm firstly. The adaptive edge detection algorithm is robust to illumination variance so that it extracts lots of edges and breakouts edges steadily in border between background and face areas. Because of strong edge detection, face area appears one or multi regions. We can merge these isolated regions using color information and get the final face area as a MBR (Minimum Bounding Rectangle) form. If the size of final face area is under or upper threshold, color sampling method in center area from input image is used to detect new face area. To evaluate the proposed method, we have experimented with 2,100 face images. A high face detection rate of 96.3% has been obtained.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.4
• /
• pp.1424-1440
• /
• 2015

It is a challenging problem to search the intended images from a large number of candidates. Content based image retrieval (CBIR) is the most promising way to tackle this problem, where the most important topic is to measure the similarity of images so as to cover the variance of shape, color, pose, illumination etc. While previous works made significant progresses, their adaption ability to dataset is not fully explored. In this paper, we propose a similarity learning method on the basis of probabilistic generative model, i.e., probabilistic latent semantic analysis (PLSA). It first derives Fisher kernel, a function over the parameters and variables, based on PLSA. Then, the parameters are determined through simultaneously maximizing the log likelihood function of PLSA and the retrieval performance over the training dataset. The main advantages of this work are twofold: (1) deriving similarity measure based on PLSA which fully exploits the data distribution and Bayes inference; (2) learning model parameters by maximizing the fitting of model to data and the retrieval performance simultaneously. The proposed method (PLSA-FK) is empirically evaluated over three datasets, and the results exhibit promising performance.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.11 no.5
• /
• pp.313-323
• /
• 2016

With the development of autonomous vehicle, there has been active research on advanced driver assistance system for road marking detection using vision sensor and 3D Laser scanner. However, vision sensor has the weak points that detection is difficult in situations involving severe illumination variance, such as at night, inside a tunnel or in a shaded area; and that processing time is long because of a large amount of data from both vision sensor and 3D Laser scanner. Accordingly, this paper proposes a road marking detection and classification method using single 2D Laser scanner. This method road marking detection and classification based on accumulation distance data and intensity data acquired through 2D Laser scanner. Experiments using a real autonomous vehicle in a real environment showed that calculation time decreased in comparison with 3D Laser scanner-based method, thus demonstrating the possibility of road marking type classification using single 2D Laser scanner.

• Journal of KIISE
• /
• v.43 no.2
• /
• pp.186-195
• /
• 2016

This paper proposes a vehicle license plate detection method in real road environments using 8 bit-MCT features and a landmark-based Adaboost method. The proposed method allows identification of the potential license plate region, and generates a saliency map that presents the license plate's location probability based on the Adaboost classification score. The candidate regions whose scores are higher than the given threshold are chosen from the saliency map. Each candidate region is adjusted by the local image variance and verified by the SVM and the histograms of the 8bit-MCT features. The proposed method achieves a detection accuracy of 85% from various road images in Korea and Europe.

• Journal of Information Processing Systems
• /
• v.14 no.6
• /
• pp.1494-1507
• /
• 2018

Intelligent human identification using face information has been the research hotspot ranging from Internet of Things (IoT) application, intelligent self-service bank, intelligent surveillance to public safety and intelligent access control. Since 2D face images are usually captured from a long distance in an unconstrained environment, to fully exploit this advantage and make human recognition appropriate for wider intelligent applications with higher security and convenience, the key difficulties here include gray scale change caused by illumination variance, occlusion caused by glasses, hair or scarf, self-occlusion and deformation caused by pose or expression variation. To conquer these, many solutions have been proposed. However, most of them only improve recognition performance under one influence factor, which still cannot meet the real face recognition scenario. In this paper we propose a multi-scale parallel convolutional neural network architecture to extract deep robust facial features with high discriminative ability. Abundant experiments are conducted on CMU-PIE, extended FERET and AR database. And the experiment results show that the proposed algorithm exhibits excellent discriminative ability compared with other existing algorithms.