基于PCNN的景象匹配和图像拼接  

Scene Matching & Image Mosaic Based on PCNN

作　　者：张绍堂[1] 杨丽云[2] 

机构地区：[1]曲靖师范学院计算机科学与工程学院,云南曲靖655011 [2]云南大学信息学院,云南昆明650091

出　　处：《计算机仿真》2009年第11期231-234,共4页Computer Simulation

摘　　要：在实际应用中,采集到的实时图像会受到噪声的干扰,或相对于基准图有一定的旋转、比例变化等。因此,要求景象匹配算法必须具有一定的容错和抗干扰能力。将脉冲神经耦合网络思想引入景象匹配中,提出了一种用PCNN提取图像熵序列来实现景象匹配的新方法,具有较好的旋转不变性,对噪声和尺度变化具有很好的鲁棒性。在PCNN模板匹配的基础上还实现了图像良好拼接,具有对准精度高、易于硬件实现的特点。仿真实验结果验证了方法的有效性。In practice, the real - time images always have the characteristics subjecte＆ to noise interference, or having rotation related to the base map and the proportion changes. Therefore, the scene matching algorithms must have a certain fault - tolerant and anti - interference capability. This paper introduces the concept of pulse coupled neural network （PCNN） into the ＂scene matching＂, and suggests a new method which implements scene matching by using PCNN for getting the vision entropy sequence. This method provides good rotation invariance as well as good robustness to the noise and the scale change. Based on the PCNN template matching, this paper shows a solution for good image mosaic which is accurate and easy to implement the hardware. The validity of the method is verified by simulation experiment.

关 键 词：脉冲神经耦合网络 特征系列 景象匹配 图像拼接 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]