Aiming　at　the　problem　that　the　existing　gray-based　template　matching　algorithm　is　not　resistant　to　rotation　and　slow　operation,　combined　with　fast　Fourier　algorithm　and　image　pyramid　search　strategy,　a　new　fast　matching　algorithm　for　angle-adaptive　grayscale　templates　was　designed.　Firstly,　the　image　is　preprocessed　to　reduce　the　noise　interference　during　matching　Secondly,　the　multi-angle　to-be-detected　gallery　is　built　by　rotating　the　sub-image　to　be　matched,　which　solves　the　problem　of　contour　artifact　caused　by　template　rotation.　Finally,　the　FFT-based　NCC　algorithm　is　used　to　measure　the　similarity.　Combined　with　the　image　pyramid　search　strategy　and　the　indented　search　step　method,　the　computational　complexity　is　reduced　and　the　robustness　is　improved.　And　this　algorithm　is　compared　with　the　traditional　anti-rotation　algorithm　during　experiments.　Experiments　show　that　the　proposed　method　achieves　good　results　when　the　angle　deviation　is　large　and　the　image　matching　probability　is　low　or　mismatched.　©2019　IEEE.