论文部分内容阅读
针对单帧红外图像中的弱小目标检测问题,提出了一种结合小波包和高阶统计量的新方法.首先,利用小波包变换对图像进行频域上的分解.然后,针对小波包树上的节点,由低到高采用基于四阶累计量的高斯判别准则合并相邻四个全高斯性小波包系数,得到图像的最优划分.接下来,将最低频带上的小波包系数和高斯性小波包系数置零来分别抑制背景杂波和噪声.最后,采用这些新的系数即可重建检测结果.实验结果表明:该方法能够稳健、有效地检测红外弱小目标.“,”Aiming to the problem of detecting small targets in a single frame infrared image, a new method combining wavelet packet with higher-order statistics (HOS) was presented. Firstly, the wavelet packet transform(WPT) was used to decompose the frequency bands of the image. Then, a criterion based on the fourth-order cumulant was utilized to merge four Gaussian wavelet packet coefficients bottom-up-in the wavelet packet tree. After that, an optimal image partition was obtained. Thirdly, the coefficient at the lowest frequency band and the Gaussian coefficients were set to zero to suppress background clutter and remove noise respectively. Finally, the detection result was reconstructed by the new coefficients. The experimental results show that the proposed method is robust and effective for small IR targets detection.