http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Mesh-Based View Interpolation Using Adaptive Node Insertion and Elimination
Sung-Ho Lee,Seung-Won Jung,Seung-Kyun Kim,Sung-Jea Ko 대한전자공학회 2010 ICEIC:International Conference on Electronics, Inf Vol.1 No.1
This paper describes a triangular mesh based novel virtual view interpolation algorithm. In the conventional mesh based view interpolation, the feature points are extracted and the disparity between images is estimated at the extracted points. Then, image warping is performed by using the estimated disparity values. In order to improve the performance of the conventional view interpolation techniques, we first insert additional feature points at the disparity discontinuity and eliminate unnecessary feature points. Then, the forward and backward warping results are combined by using an adaptive weighting factor. The experimental results show that the proposed method improves the visual quality of the interpolated image without excessively increasing the computational complexity.
Multi-scale Convolutional Neural Network 기반의 컬러 영상 가이드를 활용한 깊이 영상 초해상도 기법
박상현(Sang-Hyun Park),김준연(Joon-Yeon Kim),고성제(Sung-Jea Ko) 대한전자공학회 2020 대한전자공학회 학술대회 Vol.2020 No.8
In this paper, we propose a color guided depth map super resolution method based on multi-scale convolutional neural network. The proposed method gradually improves resolution of the depth map by using each scale information through subnetwork. To train the network, each subnetwork is supervised by the down-sampled ground truth depth map. In addition, we reduced the number of trainable parameters through the weight sharing of the subnetworks. Experimental results demonstrate that the proposed method shows better performance than the conventional method.