CAJ | 학술논문

由于物理结构的限制,单CCD图像传感器在每个像素位置上只能采集一个颜色分量,其余两个颜色分量只能通过插值的办法得到.提出一种在色差空间进行插值的算法,利用具有比例线性关系的局部颜色特性和局部协方差求得最小均方误差意义下的最优插值系数.所获取的该捕值系数具有边缘细节保持性,同时色差空间的插值又考虑了不同颜色分量间的耦合性,因而有效地抑制了图像边缘的模糊和颜色失真.实验结果表明,该算法可获得高质量的插值图像.
유우물리결구적한제,단CCD도상전감기재매개상소위치상지능채집일개안색분량,기여량개안색분량지능통과삽치적판법득도.제출일충재색차공간진행삽치적산법,이용구유비례선성관계적국부안색특성화국부협방차구득최소균방오차의의하적최우삽치계수.소획취적해포치계수구유변연세절보지성,동시색차공간적삽치우고필료불동안색분량간적우합성,인이유효지억제료도상변연적모호화안색실진.실험결과표명,해산법가획득고질량적삽치도상.
Due to the physical structure of a single-chip CCD image sensor, there is only a single color component at each pixel position. To recover the full color images, an image interpolation or de-mosaicking is required to reconstruct the other two color components. In this paper, a novel image interpolation algorithm in the color difference space is proposed. By using the ratio linearity property of local color and the local covariance, the optimal MMSE interpolation coefficients are obtained. These coefficients can preserve the edge details of the interpolated images. The interpolation in the color difference space takes into account the correlations between color components, so that the edge blurring and color distortion can be dramatically reduced. Compared with some of the state-of-the-art methods, experiment results demonstrate that the interpolated images using the proposed algorithm are of good quality.