To solve the problem of worse feasibility, lower resolution of continuous curve without thin beds information in stratum of traditional data pre-processing methods in through casing resistivity logging, a new data pre-processing method is designed in this paper. Abnormal values on same depth are eliminated by data cleaning theory and the sparse representing coefficients of signal are solved by optimizing algorithm and constraints through constructing observation matrix and sparse transforming matrix based on signal sparse representing theory to get the continuous through casing resistivity logging curve with equal space of depth interval. Experimental result is that abnormal values are eliminated effectively and formed logging curves by this method has good correspondence of thin beds with open-hole micronormal curves on logging data of five wells in SL oilfield. Results show that compared with traditional pre-processing methods, the method has good usability, stronger thin beds dividing ability and more greater application value. The implemented software provides powerful technical support for development and sustained and stable production in the oilfield during the application of dynamic supervising of SL oilfield.