为进行空间X射线星图分割及在X射线背景噪声中提取目标源,提出了改进的脉冲耦合神经网络图像分割算法。改进算法将图像分割成小区域,不同区域中设置不同的参数;将内部行为的调制参数设计成邻域像素灰度值分布的负耦合函数,强制降低强x射线目标源附近的强x射线背景噪声的点火频率,以提取x射线辐射点目标源。分别利用改进型PCNN模型、普通PCNN模型进行了空间X射线星图分割实验,结果表明改进型PCNN模型具有更优的分割效果。
For the purse of x-ray sky image segmentation, an algorithm based on improved pulsed coupled neural networks was presented. Firstly, the sky image was partitioned into separate sections, and different parameters were set in different sections. Then, the modulation parameter of inner action was designed as negative-coupling of adjacent pixel power, which compelled to depress the firing frequency of x-ray noise adjacent to x-ray object. The experiments show that improved PCNN segmentation algorithm is more effective than the former PCNN.