To　improve　the　recovery　performance　of　compressed　sensing　(CS)　recon-　struction,　this　paper　proposed　a　CS　reconstruction　model　based　on　neural　dynamics　optimization　algorithm　(NDOA)　with　l1-norm　minimization,　and　applied　it　to　the　two-　dimensional　(2D)　image　signals　to　verify　the　reconstruction　effectiveness.　Experiments　proved　the　NDOA　had　performed　better　in　reconstruction　than　the　orthogonal　matching　pursuit　(OMP)　and　its　improved　algorithms　considering　its　perfect　recovery.　When　the　compressed　sampling　rate　is　0:875　for　64×64　Lena　image,　the　application　of　NDOA　would　raise　the　peak　signal　to　noise　ratio　(PSNR)　by　3:6466dB　while　decrease　the　relative　error　(REOR)　by　2:21%　in　comparison　with　OMP　and　SWOMP.　It　showed　that　the　effective-　ness　of　CS　reconstruction　of　NDOA　was　evident.　At　the　same　time,　the　adoption　of　NDOA　could　realize　real-time　processing　by　parallel　computing,　which　would　effectively　solve　real-time　problems　such　as　image　processing　in　the　realm　of　high　dimensional　sparse　signals.　©　2017.