结合小波神经网络的CCD相机噪声处理方案

CCD Camera Noise Removal Technique Combined with Wavelet Neural Network

CCD相机响应功能复杂的非线性特征导致传统的线性滤波方式效果不理想.在分析CCD噪声模型的基础上,用小波神经网络非线性逼近噪声曲线,然后根据噪声参数进行分类自适应去噪.理论分析和实验结果证明该滤波方式可行,通过与传统滤波器相比较,噪声消除、边缘保留及信噪比方面均得到提高.

The complexity of the nonlinear characteristics of the CCD camera response function leads to the imperfect effect in the way of traditional linear filtering.Based on analysis of CCD noise model,wavelet neural network(WNN) was used to approach the photon transfer curve(PTC) in a nonlinear way,and classified the noise and removes the noise adaptively according to the noise parameter.It was proved feasible by theory analysis and the experiment results.Comparing it with the traditional filter,noise elimination,edge reservation and SNR were improved.