基于聚类融合的中波红外图像目标提取方法

Target extraction method of medium wave infrared image based on cluster fusion

为了提升目标提取效果、降低环境因素的干扰,提出基于聚类融合的中波红外图像目标提取方法。首先对中波红外图像实施小波阈值去噪及增强处理,加强中波红外图像的清晰度;其次采用多模态特征融合方法提取图像中目标物的对比度特征及熵特征;然后通过聚类融合多种特征实现对目标的准确提取。实验结果表明:应用该方法后,去噪后的峰值信噪比最大值可达到118 dB,且可以完成地提取出中波红外图像中的6处目标,证明本方法具有准确率高、有效性强的应用优势。

In order to improve the effect of target extraction and reduce the interference of environmental factors,a medium wave infrared image target extraction method based on clustering fusion is proposed.Firstly,wavelet threshold denoising and enhancement are applied to the medium wave infrared image to enhance the clarity of the medium wave infrared image.Secondly,the multi-modal feature fusion method is used to extract the contrast feature and entropy feature of the target in the image.Then the target can be extracted accurately by clustering and fusing multiple fea-tures.Experimental results show that the maximum SNR after denoising can reach 118 dB,and 6 targets in medium wave infrared images can be successfully extracted,which proves that the proposed method has the advantages of high accuracy and strong effectiveness.