针对红外光学系统在复杂背景下的弱小目标检测问题, 建立了基于特征整合的信息处理模型, 提出了采用视觉特征整合的弱小目标检测方法。该方法首先利用视网膜神经节细胞感受野的数学模型DOG(Different-of-Gaussian)对红外图像进行初级信息处理, 初步检测出弱小目标。而后, 分为空域和频域两个通道进行特征提取。在空域通道, 利用图像信息构造二阶微分Hessian矩阵, 通过计算其直迹与行列式进行局部极值的判定, 提取出含有弱小目标的结构分量特征; 在频域通道, 利用小波对图像频域进行二级分解, 提取出含有弱小目标的高频分量特征。最后, 将空域通道与频域通道的分量特征整合, 提取出复杂背景下的弱小目标。实验结果表明: 当虚警率为10-3时, 该方法对弱小目标的平均检测概率为95.17%。基本满足红外弱小目标检测方法的稳定可靠、精度高等要求。

To solve the problem of detecting small infrared targets ininfrared optical systems under complex backgrounds, an information processing model based on visual feature integration was established, and a dim small target detection method based on visual feature integration was proposed. First, the difference of Gaussians model of the receptive fields of retinal ganglion cells was used for processing the primary information of infrared images and initially detecting dim small targets. Then, the features containing the dim small targets were extracted by spatial and frequency channels. In the spatial channel, the second-order differential Hessian matrix was constructed with image information, and the local extremum was determined by calculating the basis truth and determinantto extract the structural component features. In the frequency domain channel, wavelets were used for decomposing the image frequency domain to extract the features of high-frequency components containing dim small targets. Finally, the spatial channel and frequency channel were integrated to extract the dim small targets from complex backgrounds. Experimental results indicate that when the false alarm rate is 10-3, the average detection probability is 95.17%. The proposed method can basically meet the requirements of stability, reliability, and high accuracy.