中文摘要：

针对红外图像通过普通小波阈值去噪不能较好地保留边缘信息的问题，提出了一种数学形态学边缘检测和小波阈值去噪相结合的方法，对摩擦副表面红外图像进行去噪，达到获得较为准确的温度场的目的。红外图像经过小波变换，在高频子带中做数学形态学边缘检测，确定边缘信息的位置，再进行阈值去噪处理。试验结果表明，相比普通小波闽值去噪方法，该方法不仅较好地保留了红外图像的边缘信息，去噪效果明显，而且改善了均方误差和峰值信噪比。该方法意在提高红外图像测温的准确性，为测量和分析摩擦温度场提供更好的技术支持，具有较高的工程应用价值。

英文摘要：

The common method of wavelet threshold denoising for infrared linage processing usu-ally distorts the edge in image. A new method, combining the mathematical morphology edge detection and the wavelet threshold denoising, was presented and applied in denoising the infrared image of friction surface. Infrared image was processed through three steps, wavelet transformation, mathe- matical morphology edge detection in the high--frequency sub--band for determination the location of edge and threshold denoising. Experimental results show that this method can preserve more edge in- formation, improve mean square deviation and peak signal--to--noise ratio. The intention of this ap- proach is to improve the accuracy of the temperature measurement of the infrared images, thus it pro-vides better technical support for measuring and analyzing friction temperature field with the high engineering values.