In this paper, a structure tensor based approach is proposed for multi-focus image fusion within the wavelet framework. Structure tensor is employed to extract local features in detail sub-bands. A nonlinear flow based on the trace of the structure tensor matrix is applied to matrix element before calculating the eigenvalues. The source data with larger eigenvalue contains more geometric features. An adaptive weight function is constructed to yield new detail coefficients of the fused image. Experimental results show that the proposed scheme improves performance compared to some related wavelet approaches.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.