中文摘要：

纹理特征是红外图像中的重要因素之一。提出了一种复杂纹理背景下的红外目标提取算法。由于均值漂移算法是一种非参数密度估计算法,在图像分割中得到了广泛应用,所以首先通过均值漂移算法对红外图像进行平滑处理,以消除噪声对图像质量的影响;然后对平滑后的红外图像的像素进行均值漂移图像分割,并结合8邻域差值聚类法提取出红外目标前景信息。与传统的纹理处理方法相比,该算法可剔除纹理背景而保留非纹理目标,无监督性和适应性较好,在军事领域具有一定的应用价值。

英文摘要：

An texture feature is one of the important elements in an infrared image.An algorithm for extracting infrared targets against complex texture background is proposed.First,the mean shift algorithm,a nonparametric density estimation algorithm widely used in image segmentation is used for the smooth processing of an infrared image,so as to eliminate the influence of noise on the image quality.Then,the mean shift image segmentation is implemented on the pixels in the smoothed infrared image and is used to extract the infrared target region together with an eight neighborhood difference clustering process.Compared with the traditional texture processing methods,this algorithm can remove the texture background while remaining the non-texture target.Because of better adaptability,it is of a certain value to military applications.