In order to locate and exclude the singularity induced by vibration and to evaluate the surface texture exactly,a wavelet singularity detection and filtering method is presented. Firstly,the wavelet singularity analysis is used to preprocess the sampled surface and then the 3-dimensional roughness profile is extracted by using the wavelet filter. Then, the typical evaluation parameters of roughness profile are computed and the results are compared with two kinds of samples: measured data manually added singularity and measured data with singularity originally. Experimental results show that for the first kind of data roughness parameters after singularity processing for the first kind of data with added singularity are similar to those of original surface,and the relative changes of Sa,Sz,Sq,Ssk and Sku are 0.27%、0.78%、0.51%、0.04%、1.04% respectively. The relative changes of Sa,Sz,Sq,Ssk and Sku before and after singularity processing for the first kind of data with added singularity are 1.14%、20.63%、1.95%、3.81%、8.86%. For the second kind of data the relative changes of Sa,Sz,Sq,Ssk and Sku before and after singularity processing are 6.2％、27.47％、9.54％、131.62％、53.77％respectively. The experiments prove that more reasonable evaluation results can be obtained if the singularity is excluded from the surface before the roughness profile is extracted by using the wavelet filtering.