Four-scenes comics in Japanese newspapers are composed of four related pictures and organize one story based on current social topic. Though the subject to extract person objects from such pictures and identify the same objects among different scenes is one of graphics recognition issues, it is not easy to perform this subject successfully because individual objects are deformed more tremendously than these in the ordinarily investigated graphics recognition. In this paper, we address an extraction and identification subject of person as the first step for understanding the story of four-scenes comics in Japanese newspaper. Our technique for extracting person objects in each scene is to infer person objects from the hair areas, and our idea for identifying the same person object among different scenes is to apply the distance distribution among component pixels to the distinction of person objects. This paper discusses our experimental method and shows the recognition result clearly through experiments.