As complex color shading becomes increasingly prevalent in contemporary illustration, vectorization techniques must keep up to facilitate vector processing of such images. This paper introduces a method that autonomously identifies regions with smooth color variations, classifies and reconstructs them as linear gradients, radial gradients, or solid fills as appropriate for vectorization, and then determines their geometry. We demonstrate the efficacy, accuracy, and robustness of this method, particularly in managing images characterized by intricate color shading. The outcome is a vectorized representation of the input image that preserves visual accuracy while minimising geometric complexity and improving editability