Although Spiral Architecture (SA) has many advantages in image processing and machine vision, there is no available image capture device yet to support this structure. Hence, in order to implement our theoretical results, it is necessary to construct the SA from the existing image structure, on which the traditional image representation is based. Therefore, in this paper we provide two methods, Mimic model and Pseudo model. We can construct the architecture from the rectangular pixels and thus they share the Spiral addressing mechanism. This paper then presents how Pseudo model may help to evaluate the performance of image compression on Spiral Architecture. Based on the two unique properties of Spiral Architecture, i.e. locality of light intensity and uniformity of image partitioning, possible approaches to image compression may be discovered. Our future work would focus on partitioning the image into blocks and implement compression on the transform domain.