Stereo matching is currently one of the most active research topics in domain of computer vision. A sub-line segment match method based on the wavelet transformation is presented, in order to match feature points more accurately and efficiently. It is based on that the multi-scale local maxima of wavelet transform can obtain the locations of the signal sharp variation points at different scales for transformation coefficients of a quadratic spline wavelet. A from-coarse-to-fine matching algorithm is applied to match these feature points. Sub-line segment is structured by these feature points. The points between the two endpoints of the sub-line segment are matched quickly using acceleration way. This method proposes a good solution for balance problem between match precision and match speed. The thickness of cement clinker in the cooler is measured using this method. The experiment shows this method can obtain the bed of material distributed condition of the cement clinker precisely.