Abstract: Due to the complexity of traffic environments and the high mobility of vehicles, automotive millimeter-wave synthetic aperture radar (SAR) systems face two critical challenges during imaging: motion error compensation and clutter/noise suppression. Given that existing methods typically handle autofocus and image enhancement separately, making it difficult to simultaneously address motion compensation and background suppression, this paper proposes an autofocus sparse joint optimization method for automotive millimeter-wave SAR imaging. The proposed method embeds the phase error estimation obtained from autofocus into the imaging process, constructs an autofocus imaging operator, and utilizes a two-layer iteration to achieve sparse image reconstruction. Firstly, based on the Range-Doppler Algorithm (RDA), a fast imaging and inverse imaging operator suitable for vehicular SAR is constructed. Secondly, in the outer iteration, the phase compensation term estimated by the Weighted Phase Gradient Autofocus (WPGA) method is embedded into the RDA operator to form a dynamic observation model with autofocus capability. Subsequently, in the inner iteration, image-domain sparse regularization is introduced based on data fidelity constraints, and low-amplitude unstructured clutter and noise are suppressed through soft threshold iteration. Comparative analysis using measured data demonstrates that the proposed method can simultaneously realize motion compensation and clutter/noise suppression, enhancing both the resolution and overall quality of SAR images and thereby validating its effectiveness in automotive SAR imaging.