Artificial　neuromorphic　tactile　systems　based　on　tactile　sensors　and　artificial　neurons　can　simulate　biological　perception　systems　and　achieve　the　efficient　transmission,　processing　and　utilization　of　information.　However,　limited　by　the　materials　and　structures　of　devices,　it　is　difficult　for　artificial　neuromorphic　tactile　systems　to　have　stable　perception　abilities　such　as　biological　functions,　especially　when　the　devices　are　bending　or　stretching.　In　this　work,　we　propose　a　performance-stable　artificial　tactile　neuron,　which　integrates　a　stretch-insensitive　triboelectric　nanogenerator　(TENG)　with　an　artificial　neuron　in　a　single　device.　The　stable　contact　triboelectric　output　could　stimulate　a　neuron　and　affect　its　firing　time.　Based　on　this　principle,　we　established　a　64　x　64　neuromorphic　tactile　matrix,　which　processed　the　triboelectric　signal　generated　by　touch　through　the　integrate-and-fire　neuron　and　simulated　the　pressure　trajectory　and　texture　of　tactile　perception　based　on　the　firing　time　classification.　This　provides　an　effective　strategy　for　the　future　simulation　of　biological　skin　perception　and　fast　classification　and　recognition　of　information.　A　performance-stable　tactile　neuron　is　developed,　which　integrates　a　stretch-insensitive　triboelectric　nanogenerator　with　an　artificial　neuron　in　a　single　device,　and　a　64　x　64　neuromorphic　tactile　matrix　is　established　to　process　touch　signals.