煤矿工业电视图象的小波神经网络压缩  

Compression of Wavelet Neuron Network for Coal Industrial Television Picture

作　　者：王正友[1] 于洪珍[1] 

机构地区：[1]中国矿业大学信电学院,江苏徐州221008

出　　处：《矿山机械》2001年第1期59-60,共2页Mining & Processing Equipment

摘　　要：首先将煤矿工业电视图象分为相对静止区和相对运动区两部分．其次对相对运动区通过小波变换将图象信号分解为若干不同空间分辨率、频率特性和方向特性的子图象信号，然后利用基于ＢＰ算法的三层前向神经网络构造的非线性预测器，进行压缩编码传输，最后和单独处理的相对静止区图象合成输出，收到了较好的效果。The television picture is divided in this article into two sections: relative shoe zone and relative movable zone. For the relative movable zone, the picture signal can be decomposed by wavelet change into sub-picture signals such as different space resolution, frequency character and directional character, then the compressed codes is transmitted by a nonlinear predictor With three-layers forward neuron network on the basis of BP algorithm, Finally, it is combined with the picture in the relative static zone, which is treated serparately, for transmission. Good effect has been procured by the use of this method.

关 键 词：煤矿 工业电视 神经网络 小波变换 视频信号 低频子带 

分 类 号：TD655[矿业工程—矿山机电]