强噪声条件下激光光斑图像预处理方法研究

为了研究强噪声光斑图像的抑噪方法，采用基于小波阈值法和形态学滤波的级联算法及小波域中值滤波法进行预处理，以信噪比、均方根误差和光强分布作为评估标准，并将处理效果与其他方法对比。结果表明：小波域中值滤波处理效果优于空域中值滤波；级联法的抑噪和光斑特征恢复效果更优于传统方法，其中对-2.0843dB的原低信噪比图像处理后的信噪比、均方根误差分别约为小波变换和空域中值滤波法的1.34、0.81倍和3.14、0.50倍；且原图像信噪比越低，其处理效果相对于传统方法的优越性越明显。

Pretreatment for laser spot image in strong noise

To investigate the de-noise ways of the strongly polluted laser spot image, a wavelet domain median filter and a combined filter based on wavelet-threshold and mathematics morphology were applied to inhibit the strong noise of laser spot images; the image Signal to Noise Ratio (SNR) and the Root-Mean-Square Error (RMSE) and the image gray surface chart were selected to estimate the de-noising effect of the images. Compared with other image filters, experimental results show that the de-noising effect of the Wavelet domain median filter is better than spatial domain median filter; when image noise reduction is carried out with the continuum method, the improved SNR is about 1.34 times of the wavelet transform algorithm and 3.14 times of the median filter; the RMSE is about 0.81 times of the wavelet transform algorithm and 0.50 times of the median filter; moreover, when the SNR of the former laser spot image is lower, the advantage of inhibiting noise of this method is more obvious.