In the area of multiscale analysis of signals, including images, the wavelet transform is one of the most attractive and powerful tool due to its ability to focus on signals structures at different scales. In this work we show how to use the modulus maxima of the wavelet transform at different scales for multiscale singularity detection as well as for computation of multifractals properties. The proposed algorithm for Dq computation is applied to dye stained images of a crop field of 2x2 meters. A comparison is stablished with previous works where multifractal analysis is applied to quantify the self-similarity properties of images of dye-stained flow paths through the generalized dimension (Dq) of different images using a box counting technique. The image being analyzed is divided into boxes of variable size "r" which acts like an scale parameter for multiscale image analysis.