本研究基于文本和图像的综合理解模式（Wolfgang Schnotz和Maria Bannert）以及符号学模型（Roth）对Martinec和Salway提出的图像-文本关系系统中的图文地位关系框架提出创新性修订，修订后的框架与物理教科书适配性较强，具有实践意义。借用SPSS软件采用KAPPA检验方法对创新后的框架进行信度检验，统计分析表明框架信度高。依据框架对研究对象R版五版教科书（1957年版、1979年版、1991年版、2006年版、2019年版）的圆周运动与万有引力章节，电磁感应章节进行编码处理。分析不同版本教材图文地位关系分布比例，分析不同版本同一模块内容地位关系分布比例的特点。依据认知心理学基础理论以及前人提出的教科书插图设计原则和建议等理论基础，对数据结果进行分析得到结论与优化建议。《普通高中物理课程标准》对教材物理形态的要求，教材文字叙述要科学、准确、精炼，图像的选择要与教学目的、教学内容紧密相关，具有科学性、时代性。分析结果表明现有R版2019版中部分插图与文本存在不必要的要素重叠，部分文本描述可精简。

This study proposed an innovative revision of the image-text status system proposed by Martinec and Salway based on the integrated understanding model of text and picture (Wolfgang Schnotz and Maria Bannert) and Semiotic model(Roth). The revised framework fits well with physics textbook and has practical significance. Using KAPPA test method in SPSS software to test the reliability of the innovative framework, statistical analysis shows that the reliability of the framework is high. According to the framework, the chapter of circular motion and universal gravitation and the chapter of electromagnetic induction of R edition (1957 edition, 1979 edition, 1991 edition, 2006 edition and 2019 edition) were coded. This paper analyzes the distribution proportion of the  image-text relationship of status in different editions and reflects on the evolution of the distribution proportion of the status relation between the content in the same module in different editions. Based on the basic theory of cognitive psychology as well as the principles and suggestions of textbook illustration design proposed by predecessors, the data results are analyzed and conclusions and optimization suggestions are obtained. The physical form of teaching materials is required by the Curriculum Standard of Physics for Ordinary Senior High schools. The text description of teaching materials should be scientific, accurate and refined, and the selection of pictures should be closely related to the teaching purpose and content, with scientific and contemporary characteristics. The analysis results show that some of the illustrations in the existing R edition 2019 have unnecessary overlap with the text, and some text descriptions can be simplified.