Abstract: Orthogonal Haar transform (OHT) template matching algorithm uses strip sum instead of integral image to achieve better performance in 2D image process, however, it requires a standard template that has the same height and width with the power of 2. To solve this problem in OHT algorithm, the authors propose a quasi Haar transform (QHT) template matching algorithm, which uses tree division strategy to accelerate nonstandard template matching process. The QHT algorithm is applicable to both cases of standard and nonstandard templates. For standard templates, experimental results show that the QHT algorithm can be faster than the OHT algorithm in low noise levels.