Since twenty years wavelet transform was recognized as a privileged tool for analysis of the fractals objects. We exploited the self-similarity of the wavelet transform to detect singularity which is a fractal signal characterization .In a first part, we use the wavelet transform modulus maxima lines (WTMM) as a tool for analysis of synthetics fractals signals. In second part we applied this technique to the data of wells located in the Algerian Sahara, we proposed then an automatic algorithm of segmentation which is applied thereafter to a simple resolution well-logs data another with a high resolution. We demonstrate the potentialities of the method in the segmentation of different geological formations. We finalize this work by planning a multilayer perceptron neuronal machine. We used the precedents results as information able to detect the lithology and the nature of the pores fluid . Keywords: Wavelet transform, WTMM, Segmentation, simple resolution, High resolution.