结合区域生长与颜色特性的迷彩伪装设计

Camouflage Design by Combining Region Growing and Color Similarity

为增强伪装效果,缩小目标与背景的差别,提出了颜色特性与区域生长相结合的数字迷彩生成算法。算法由主色提取与目标伪装两部分组成。首先,由用户选定待伪装目标区,根据归一化颜色相似性差最小准则,应用均值聚类方法优选出背景主色;然后,利用自适应区域生长法对图像目标区域进行分割,按照颜色相似性准则,选取背景主色对目标区块进行填充;最后,采用数学形态学滤波组合进行修正以完成目标区伪装。实验表明该算法伪装效果好,与其他迷彩方案相比,伪装目标与背景颜色相似性提高约40%。此外,迷彩设计无需颜色空间转换,计算开销低。该伪装方案可应用于军事目标迷彩伪装设计,有效保护特定军事目标。

To improve camouflage effect and reduce the difference between the target and background, a digital camouflage generation algorithm combining color feature and region growing is proposed. The scheme consists of dominant color extraction and target camouflage. The target region for camouflage is first selected, and background dominant colors are obtained by using K-mean cluster in terms of minimum prin- ciple of the difference of normalized color similarity. And then an adaptive region growing is used to seg- ment the target area and the appropriate background dominant colors selected by the color similarity princi- ple are used to fill the target area. Finally, mathematical morphology filtering operations are employed to modify the camouflage target and generate the camouflage image. Experimental results show that the pro- posed camouflage generation algorithm has good camouflage effect. The color similarity of camouflaged tar- get and background is 40% higher than that of other camouflage design algorithms. Moreover,it doesn＇ t need color space transform and its computation cost is low. The camouflage generation scheme can be used to the camouflage design of military targets and protect them.