Multi-Dimensional Hash Generation for Data Falsification Detection

The patent transitions from one-dimensional hash to multi-dimensional hash by generating multiple reference hash values (first, second, third reference hash values) from different feature amounts extracted through different preprocessing processes. This dimensional expansion enables more comprehensive falsification detection while maintaining systematic complexity management through structured generation and comparison of multiple hash dimensions.

The patent segments the hash generation process into distinct preprocessing stages (first, second, third preprocessing processes) that extract different feature amounts from the same input data. Each preprocessing process generates specific feature amounts that are then quantized into separate reference hash values. This segmentation allows robust multi-dimensional hash generation while managing complexity through modular, systematic processing stages.

The patent performs preliminary extraction of multiple different feature amounts through different preprocessing processes before quantization. By preparing these feature amounts in advance through systematic preprocessing, the system enables sensitive content change detection across multiple dimensions while optimizing processing time through efficient feature extraction and quantization pipelines.