The　resolution　level　of　face　images　is　one　of　the　key　factors　affecting　the　performance　of　face　recognition　algorithms.　Face　recognition　under　low　resolution　conditions　has　always　been　a　challenging　research　topic　in　related　fields.　Using　super-resolution　restoration　technology　to　improve　the　spatial　resolution　of　the　face　to　be　recognized　and　reconstruct　its　high-resolution　information　is　an　effective　way　to　improve　the　performance　of　the　algorithm.　However,　the　traditional　image　super-resolution　restoration　algorithm　generally　has　problems　such　as　high　computational　complexity　and　difficulty　in　training,　which　restricts　its　application　in　the　actual　face　recognition　system.　Therefore,　this　paper　proposes　a　l　ow-resolution　face　image　super-resolution　restoration　algorithm　based　on　simplified　VGG　network.　Firstly,　based　on　the　degradation　process　of　low-resolution　face　images,　a　set　of　face　image　samples　corresponding　to　high　and　low　resolution　based　on　prior　knowledge　is　constructed,　and　a　streamlined　6-layer　VGG　network　is　designed　to　learn　between　high　and　low　resolution　images.　The　mapping　relationship　is　finally　achieved　by　deconvolution　amplification　to　achieve　the　super-resolution　restoration　process　of　the　image.　The　common　LFW　and　ORL　data　sets　are　used　to　test　and　analyze　the　superresolution　restoration　effect　of　the　algorithm　and　its　impact　on　the　face　recognition　algorithm.　The　experimental　results　show　that　the　proposed　algorithm　is　superior　to　the　classical　SRCNN　algorithm　in　super-resolution　performance.　When　applied　to　the　face　recognition　algorithm　based　on　Lenet-5,　its　recognition　performance　is　significantly　improved.　©　2019　Association　for　Computing　Machinery.