摘要
研究可供夜视图像进行色彩传递的自然彩色源图选择算法。利用Gabor滤波器能够模拟生物视觉的特性以及夜视图像的纹理特征,对结合Gabor滤波器和灰度共生矩阵源图检索算法做出了改进,提出了一种结合Gabor滤波器和灰度梯度共生矩阵的源图选择算法。最后对这两种算法和亮度梯度二次采样算法进行了比较,实验结果证明,改进的算法远远优于其他两个,减少了图像色彩传递应用过程的人工干预程度,大大提高了自动化水平。
This paper deals with the problem of selecting a color source image to automatically colorize a night vision image by transferring color from the source image to the night vision image. Based on the characteristics of creature-vision simulation and the texture features of image, we have improved the algorithm of combing Gabor filters and gray concurrence matrix by replacing gray concurrence matrix with gray-gradient concurrence matrix. We propose an algorithm of combining Gabor filters and gray-gradient concurrence matrix. Then we compare these two algorithms and another source image selecting algorithm, which is luminance and gradient subsampling. We conclude that the method we propose in this paper remarkably enhance its automation efficiency in the selection of source image and is much more plausible than the other two algorithms.
出处
《激光与红外》
CAS
CSCD
北大核心
2009年第2期223-226,共4页
Laser & Infrared
基金
国家自然科学基金(No.60502042)
上海市启明星基金(No.06QA14003)资助
关键词
夜视图像
色彩传递
图像检索
lαβ空间
纹理特征
GABOR小波
灰度梯度共生矩阵
night vision image
color transferring
image retrieval
lab color space
texture feature
Gabor wavelet
graylevel-gradient co-occurrence matrix