This study proposes a novel approach for robust 2D shape recognition through ellipse-fit alignment. The method aligns shapes using an ellipse fitting algorithm, addressing issues of rotation, translation, and scaling commonly observed in shape databases. Following alignment, shapes are classified using various similarity measures. Experimental results on the Kimia and TARI databases demonstrate the effectiveness of the proposed method. In the Kimia database, metrics such as Threat Score, Accuracy, F1 Score, Matthews Correlation Coefficient, Fowlkes-Mallows index, and Vote achieved perfect recognition rates. The code and databases can be accessed via https://doi.org/10.5281/zenodo.14423922. While the TARI database yielded slightly lower recognition rates, with Accuracy and Markedness contributing to an 85.5% recognition rate, the results highlight the potential and limitations of the approach. The study introduces a significant advancement in 2D shape recognition by combining ellipse fitting, affine transformation, and similarity measurement, offering a robust solution for applications in computer vision and image retrieval.