The paper presents a novel approach to shape matching and object recognition using shape contexts. The method involves solving the correspondence problem between points on two shapes, estimating an aligning transform, and computing the dissimilarity between the shapes. The shape context, a descriptor that captures the distribution of points relative to a reference point, is used to find correspondences between points on two shapes. Regularized thin-plate splines are used to estimate the transformation that aligns the shapes. The dissimilarity between shapes is computed as a sum of matching errors and the magnitude of the aligning transform. The approach is evaluated on various datasets, including handwritten digits, silhouettes, trademarks, and 3D objects, demonstrating its effectiveness in object recognition and silhouette matching. The authors also discuss the robustness of the method to deformations, noise, and outliers, and present a technique for selecting the number of stored views for each object category based on visual complexity.The paper presents a novel approach to shape matching and object recognition using shape contexts. The method involves solving the correspondence problem between points on two shapes, estimating an aligning transform, and computing the dissimilarity between the shapes. The shape context, a descriptor that captures the distribution of points relative to a reference point, is used to find correspondences between points on two shapes. Regularized thin-plate splines are used to estimate the transformation that aligns the shapes. The dissimilarity between shapes is computed as a sum of matching errors and the magnitude of the aligning transform. The approach is evaluated on various datasets, including handwritten digits, silhouettes, trademarks, and 3D objects, demonstrating its effectiveness in object recognition and silhouette matching. The authors also discuss the robustness of the method to deformations, noise, and outliers, and present a technique for selecting the number of stored views for each object category based on visual complexity.