Abstract: This paper proposes a sea-aero targets detection approach based on fractal image coding. Code the image with fast no search scheme after smooth filter, use root mean square error of each range block to estimate if there is man-made object in the block，and mark those blocks covered by artificial object, connect all the marked blocks nearby and thus obtain the whole target region. Experimental results show that the method is very effective to sea and sky background image, and it has strong noise restraining ability and is fast enough to support real time detection.

Key words: Fractal image coding, Sea-aero targets detection, Root mean square error

摘要： 提出了一种基于分形图像编码的海空目标检测方法。对图像平滑滤波后，进行快速无搜索编码，利用编码过程中各值域块的均方根误差来判断该块是否包含人造目标，标记包含人造目标的值域块，并将邻近的标记块连接起来，就可以得到完整的目标区域。实验结果表明，该方法对海空背景下的目标检测具有较高的准确率，且抗噪能力强，处理速度快，能够满足实时处理的要求。

关键词: 分形图像编码, 海空目标检测, 均方根误差

CLC Number: 

• TP391.41