Aiming at the problem that it is difficult to directly hyperdimensionally encode images of larger sizes in hyperdimensional computing and the complexity of traditional hyperdimensional coding methods is relatively high, a hyperdimensional computing classification method based on structured density is proposed. This method first uses the histogram of oriented gradients to extract the effective features of the image target and uses principal component analysis to reduce the dimension of the extracted image target features. Secondly, the computational complexity of hyperdimensional coding is reduced through structured density hyperdimensional coding. Finally, the category to which the target to be tested belongs is determined by Hamming distance. Experimental results show that this method extracts the effective features of the target in the image through traditional feature extraction methods, simplifying the difficulty of direct hyperdimensional coding. At the same time, compared with traditional hyperdimensional coding, the proposed hyperdimensional coding method reduces the computational complexity while maintaining the classification accuracy and improves the efficiency of image classification and recognition.