基于小波提升框架及小波能量的红外弱目标检测方法

针对图像序列中红外弱小目标的检测与跟踪中遇到的难点，即信噪比低、帧间相对位移小等问题，提出一种基于小波提升框架及小波能量的目标检测算法。新方法首先对隔帧融合生成的图像进行小波提升变换，然后通过计算水平及垂直方向上小波能量并选取合适的阈值将目标提取出来，再结合目标的帧间相关性实现准确跟踪。由于新方法联合考虑了帧间信息及单帧的灰度信息，因而检测结果比常规方法更为准确可靠。试验结果表明，新方法对信噪比小于2、帧间位移小于一个像素的红外斑点目标的检测与跟踪十分有效，并且实时性很好。

Dim Target Detection Method Based on Lifting Scheme and Wavelet Energy

To solve the difficulties in dim target detection, such as low SNR and small-scale displacement, a new automatic target detection algorithm based on lifting scheme and wavelet energy is proposed. First of all, Lifting scheme is used in the fused image of interval frame. Then by computing the wavelet energy both in horizontal and vertical directions, the target can be separated from background through appropriate selecting threshold value, and detected definitely through relativity of frames. The information of multiple displacement and gray-level is sufficiently used in the new algorithm. Simulation experiments show that by using this algorithm, dim target can be detected under ocean and sky natural background with SNR less than 2, continuous displacement less than 1 pixel. The experiments are also shows efficient and real-time.