Indoor panorama image brings great convenience to our life. For example, it can provide panoramic view of museums, shopping malls, hospitals and other public places, and provide the localization and navigation services based on panoramic maps. The basic of those services is the high-quality indoor panoramic images, which with invisible parallaxes and artifacts, no ghosting, rich in details and rich background texture. However, for indoor panoramic images, there exist great photometric inconsistencies and geometric misalignment due to illumination variations and the difference of camera projection centers. In addition, there also exist many dynamic objects in indoor scenes. To solve those problems, we focused our studies on the following four aspects: 1) The generation of High-Dynamic Range (HDR) images in dynamic scenes by fusing many Low Dynamic Range images captured with different exposure settings; 2) The image rectification based on the non-rigid model constructed by optical flows; 3) The optimal seamline detection and dynamic object compensation based on the graph cuts energy minimization algorithm; 4) The color adjustment and image blending to guarantee the color consistent in the final panorama images. This project aims to produce high-quality indoor panorama images and serve the needs of society and the people's life.