With increased use of smartphones and social networking services, the amount of unstructured image data has been increasing significantly. A high-speed search method is required to use such data for big data analysis. Holographic optical correlation is a promising technique for this purpose. This study introduces a recording method that realizes speedup in an optical correlator. High-density recording is effective for achieving not only speedup, but also high capacity. However, the high-density recording by shortening the shift amount in shift-multiplexed recording results in a large error in the calculated correlation rate owing to crosstalk from adjacent holograms. To suppress this crosstalk, we propose a low-correlation data interleaving method that rearranges the recording order of image data based on the correlation rate with the last recorded image data. We analytically and experimentally confirmed that the proposed method improves the recording density and the correlation speed to 1.33 times.