摘要
利用BP神经网络进行红外焦平面阵列的非均匀性校正,容易产生校正精度低、收敛速度慢及边缘模糊等缺点。文中提出一种基于中值滤波和改进网络权值的BP神经网络算法。对于非均匀性很大的红外图像,先对其进行中值滤波预处理,然后再在权值迭代过程中加入动量项改正来避免图像边缘模糊。实验结果表明改进后的BP神经网络算法更适合边缘细节丰富的红外图像。
Traditional neural network non-uniformity correction algorithm leads to a lower correction precision, lower convergence speed and slur edge. A BP neural network method for correcting the nonuniformity of an infrared focal plane array （IRFPA）based on median filtering and weight changing is proposed. For infrared image with high non-uniformity, the media filtering is first used to preprocess the high noise and then momentum part is added to avoid slurring image edge adaptively. The experimental result shows that the improved algorithm is more adaptable for the infrared images with abundance edge details.
出处
《测绘工程》
CSCD
2013年第3期24-27,共4页
Engineering of Surveying and Mapping
关键词
红外焦平面阵列
非均匀性校正
神经网络
网络权值
infrared focal plane array
neural network non-uniformity correction
neural network
network weight
