夜视图像彩色融合中基于谱残差的显著目标增强算法

Algorithm for Enhancing Salient Objects in Color Fusion of Night Vision Images Based on Spectral Residual

根据人类视觉系统的特点,提出了一种基于谱残差、可在夜视图像彩色融合中的增强显著目标的算法。该方法在彩色融合过程中,利用谱残差确定显著目标范围,用5×5的像素大小窗口对显著目标进行定位。当像素邻域均值大于图像整体均值时,此像素位于显著目标,用红色凸显目标可达到增强彩色融合图像的目的。实验结果表明,该方法具有较强的稳定性和实用性,增强显著目标的效果较好。

According to the characteristics of human visual systems, an algorithm for enhancing the salient objects in color fusion of night vision images based on spectral residual is proposed. In the color fusion process, the range of the salient object is determined by the spectral residual and the salient object is located with a window of 5x5 pixels. When the mean value in this window is greater than the global mean value, the pixels are located on the salient object. Then, the salient object is highlighted in red color so as to enhance the color fusion image. The experimental result shows that the new algorithm is stable and practical and is more effective in the enhancement of salient objects.