In recent years, software security issues have been frequent, and vulnerabilities inherited from third-party code have been troubling developers. Effective similarity detection of app code has become the key to ensuring stable system operation. This article addresses the shortcomings of conventional detection methods such as high feature dimensions and low detection efficiency. For the first time, editing distance is introduced into the similarity comparison of binary instruction sets. A new intelligent fusion terminal binary ELF file similarity determination method based on the combination of OpCode code segment editing distance and graph similarity of all common subsequences is proposed, and a quantization formula for the adjusted similarity model is proposed. It is of great significance to carry out homology or similarity analysis of fusion terminal apps, evaluate quantitative indicators of source code similarity, and solve security risks of fusion terminals!