The invention relates to caption rendering technology, belonging to television program production field, particularly to a method for defining the pixel area of a pixel space mask matrix in caption rendering based on a pixel space mask matrix model. The method is as follows: traversing the vector contours of the polygons of the caption objects, traversing each side of the N closed polygons of thecaption objects, calculating the contribution value of each side to the areas of all the pixels which pass the side, and accumulating the pixel areas (Area) of the same pixel. In this way, desired result can be obtained. As for caption objects with various original contour shapes, the method can rationally and accurately calculate the area of the polygon made up by all the contour boundaries passing one pixel, thereby providing a premises and a basis for defining the pixel type and for anti-aliasing treatment of the boundary pixels and ensuring the feasibility of implementation of the captionrendering method based on the pixel space mask matrix model.