A locally adaptive edge-preserving algorithm for image interpolation

ALZ: Adaptive Learning for Zooming Digital Images

“…The thresholds used during the local gradient analysis and the value of the factor "k" used in the equation (2) have been experimentally derived. Their values are: T1=66, T2=22, and k=1.…”

“…There is a set of them that use the gradient information in order to apply the best mask to the processed pixel. For example, Rodrigues et al [2] have proposed an adaptive edgepreserving algorithm which improves the level of the details and edges in the zoomed image by using local thresholds computation. Fu et al [3] presents a unified bidirectional flow process, where an inverse diffusion is performed to enhance edges along the normal directions to the edges, while a normal diffusion is done with respect to the other directions.…”

Modified cubic convolution scaler for edge-directed nonuniform data

“…차원 convolution [12] , 그리고 기존 방법들 [13], [16]- [18] (a) Cubic convolution [6] (b) 2D cubic convolution [12] (c) Proposed scheme 그림 7. 5.5배로 확대한 결과 영상("barbara"영상의 일부) Fig.…”

“…Reichenbach와 Geng [12] 은 2 차원 cubic convolution을 전개하고 convolution에 사용되 는 kernel parameter의 이상적인 값을 찾아냈다. 한편, edge 방향정보를 이용한 많은 보간 기술들이 제안 되었다 [13]- [18] .…”

“…그러나 edge 방향을 고려한 기존 연구들에서 scaler는 bilinear 방법을 사 용하였고 [13], [15]- [18] , scaling 비율도 2의 정수단위 지수승으 로 제한되어 사용되었다 [15]- [18] .…”