• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.42 no.6
• /
• pp.75-82
• /
• 2005

The robust feature vector selection method for the multiple frontal face detection is proposed in this paper. The proposed feature vector for the training and classification are integrated by means, amplitude projections, and its 1D Harr wavelet of the input image. And the statistical modeling is performed both for face and nonface classes. Finally, the estimated probability density functions (PDFs) are applied for the detection of multiple frontal faces in the still image. The proposed method can handle multiple faces, partially occluded faces, and slightly posed-angle faces. And also the proposed method is very effective for low quality face images. Experimental results show that detection rate of the propose method is $98.3\%$ with three false detections on the testing data, SET3 which have 227 faces in 80 images.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.1
• /
• pp.103-110
• /
• 2011

A widely used 3D face reconstruction method, structure from motion(SfM), shows robust performance when frontal, left, and right face images are used. However, this method cannot reconstruct a self-occluded facial part correctly when only one side view face images are used because only partial facial feature points can be used in this case. In order to solve the problem, the proposed method exploit a constrain that is bilateral symmetry of human faces in order to generate bilateral facial feature points and use both input facial feature points and generated facial feature points to reconstruct a 3D face. For quantitative evaluation of the proposed method, 3D faces were obtained from a 3D face scanner and compared with the reconstructed 3D faces. The experimental results show that the proposed 3D face reconstruction method based on both facial feature points outperforms the previous 3D face reconstruction method based on only partial facial feature points.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10b
• /
• pp.688-690
• /
• 2004

In this paper, we propose a new method of removing glasses from human frontal facial images. We first detect the regions occluded by the glasses, and generate a natural looking facial image without glasses by recursive error compensation using PCA reconstruction. The resulting image has no trace of the glasses frame, nor of the reflection and shade caused by the glasses. The experimental results show that the proposed method provides an effective solution to the problem of glasses occlusion, and we believe that this method can also be used to enhance the performance of face recognition systems.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2010.11a
• /
• pp.70-71
• /
• 2010

요즘 감시카메라 시장은 전 세계적으로 이슈가 되고 있다. 감시카메라는 적은 인원으로 많은 장소를 한 눈에 감시할 수 있고, 문제가 발생했을 때, 녹화 저장된 영상을 통해 그 상황을 다시 볼 수 있다. 이렇게 다양한 기능과 편리함으로 우리에게 도움을 주는 감시카메라이지만, 마스크나 선글라스, 또는 여러 가지 잡음에 의해 얼굴 영상의 부분이 훼손되는 상황에서는 신원 확인을 하기가 어렵다. 때문에 가려진 얼굴을 제대로 인식하기 위해 영상 처리 분야에서 얼굴의 가려진 영역을 찾아 그 부분을 재구성하고자 하는 연구가 활발히 진행되고 있다. 이에 본 논문은 기존 PCA 방법을 이용하여 가려진 영역을 찾아내고, PCA를 반복적으로 사용하여 재구성하는 대신 상관관계를 이용하여 얼굴의 가려진 영역을 자동적으로 검출하고 복원하는 방법을 제안한다. 본 논문에서는 두 눈의 중심이 고정되어 있는 BioID 데이터로 상관계수를 구하고 얼굴의 특정 부분을 임의로 가려 실험을 수행하였다. 제안된 방법의 결과는 PCA 방법으로 수행한 결과와 함께 비교되어 원본 영상과의 오류 값이 더 작게 나오는 것을 확인할 수 있었다.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.28.3-28
• /
• 2001

This paper proposes a new algorithm that detects and separates the occluding and occluded objects in a 2D image. An input image is represented by the attributed graph where a node corresponds to a surface and an arc connecting two nodes describes the adjacency of the nodes in the image. Each end of arc is weighted by relation value which tells the number of edges connected to the surface represented by the node in the opposite side of the arc. In attributed graph homogeneous nodes pertained to the same object always construct one of three special patterns which can be simply classified by comparison of relation values of the arcs. The experimental results have shown that the proposed algorithm efficiently separates the objects ...

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.5
• /
• pp.191-198
• /
• 2013

This paper proposes detection of view reversal in a stereo video using depth map and motion vector information. We obtain a depth map by using a stereo matching and divide the input image into foreground and background. Next, we obtain a motion vector field by using a motion estimation. In general, an occluded region is in background when foreground goes toward the adjacent background or the background goes toward the adjacent foreground. But, we will face with the change of foreground and background because their depths also change when view reversal occurs. Therefore, we can detect the view reversal in stereo videos by using the observation that the foreground goes toward the adjacent background or the background goes toward the adjacent foreground. The experimental results show that the proposed algorithm achieves good detection rate when the background region is sufficiently occluded by the moving foreground.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.3
• /
• pp.35-49
• /
• 2004

This paper proposes a new glasses removal method from color frontal facial image to generate gray glassless facial image. The proposed method is based on recursive PCA reconstruction. For the generation of glassless images, the occluded region by glasses should be found, and a good reconstructed image to compensate with should be obtained. The recursive PCA reconstruction Provides us with both of them simultaneously, and finally produces glassless facial images. This paper shows the effectiveness of the proposed method by some experimental results. We believe that this method can be applied to removing other type of occlusion than the glasses with some modification and enhancing the performance of a face recognition system.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.38 no.1
• /
• pp.54-68
• /
• 2001

This paper proposes a new algorithm that detects and separates the occluding and occluded objects in a 2D image. An input image is represented by the attributed graph where a node corresponds to a surface and an arc connecting two nodes describes the adjacency of the nodes in the image. Each end of arc is weighted by relation value which tells the number of edges connected to the surface represented by the node in the opposite side of the arc. In attributed graph, homogeneous nodes pertained to a same object always construct one of three special patterns which can be simply classified by comparison of relation values of the arcs. The experimental results have shown that the proposed algorithm efficiently separates the objects overlapped arbitrarily, and that this approach of separating objects before matching operation reduces the matching time significantly by simplifying the matching problem of overlapped objects as the one of individual single object.