谱色测温系统中图像噪声会影响温度计算的精度,需采用数字图像处理方法进行降噪。针对系统中两种工作模式的不同特点,需要设计不同的降噪方案。单帧分析模式下重点研究了傅里叶变换法去除条带噪声,解决了智能寻找峰值点的难题。连续采集模式下提出了多帧加权平均的算法,能同时有效地降低两种噪声。
As the image noise affects in the temperature accuracy in the primary spectrum pyrometry system, the digital image processing methods are needed for denoising. Different denoising programmes are proposed for the two working mode. In single-frame analysis mode, it focus on the Fourier transform method to remove the stripping noise and the searching for the peak point. In continuous acquisition mode, the multi-frame weighted average method is proposed to reduce both kind of noises at the same time.