Imagery registration and rectification is a process of transforming different sets of data into one coordinate system. A new model, i.e., the generalized-line-based iterative transformation model (GLBITM), is proposed by integrating the line-based transformation model (LBTM) and generalized point photogrammetry (GPP). First, the initial value of an affine transformation is acquired by LBTM. Then, on the basis of ground control lines (GCLs), not ground control points, the linear feature adjustment model with GPP is extended to a quadratic polynomial model and utilized to iteratively solve transformation coefficients. This process eliminates the translation amount and recalculates the scale and rotation coefficients. The authors suggest an iterative method with variable weights that is based on posterior variance estimation to improve quality control. A significant characteristic of the GLBITM is that the two endpoints of the corresponding GCLs are not necessarily conjugate points. The GLBITM integrates the advantages of the LBTM and GPP and avoids their respective shortfalls. Finally, this experiment verifies that the GLBITM gives correct, robust, and effective results that can be applied in high-resolution satellite imagery processing of multiple sensors, angles, and resolutions.