This paper introduces two new regional shape descriptors for 3D object recognition in noisy and cluttered scenes: 3D shape contexts and harmonic shape contexts. These descriptors are compared to the existing spin image descriptor. The study evaluates their performance on recognizing vehicles in range scans using a database of 56 cars. The results show that shape context-based descriptors outperform spin images in noisy scenes, while 3D shape contexts outperform others in cluttered scenes. The paper also discusses the use of locality-sensitive hashing to speed up search processes for 3D shape contexts, maintaining accuracy while reducing computational cost. The experiments demonstrate that 3D shape contexts achieve high recognition rates, even in challenging conditions, and that harmonic shape contexts perform poorly. The study highlights the effectiveness of regional point descriptors in balancing global and local approaches for robust 3D object recognition.This paper introduces two new regional shape descriptors for 3D object recognition in noisy and cluttered scenes: 3D shape contexts and harmonic shape contexts. These descriptors are compared to the existing spin image descriptor. The study evaluates their performance on recognizing vehicles in range scans using a database of 56 cars. The results show that shape context-based descriptors outperform spin images in noisy scenes, while 3D shape contexts outperform others in cluttered scenes. The paper also discusses the use of locality-sensitive hashing to speed up search processes for 3D shape contexts, maintaining accuracy while reducing computational cost. The experiments demonstrate that 3D shape contexts achieve high recognition rates, even in challenging conditions, and that harmonic shape contexts perform poorly. The study highlights the effectiveness of regional point descriptors in balancing global and local approaches for robust 3D object recognition.