Abstract: Applying traditional two-dimensional Otsu threshold algorithm in Gastric Adenocarcinoma Cell Image is not ideal,it is long-paying computation and has too much noise.So this paper presents a improved two-dimensional Otsu threshold automatic segmentation algorithm.The algorithm gets the optimization threshold value approximately by using new area partition method.And importing relaxant variable, enlarging the searching step of threshold ,and then gaining optimization threshold value at last.The experimental result has demonstrated that the algorithm can overcome more noise and pay less computational time than the traditional one.