The somatosensory system is the inspiration for diverse neuromorphic technologies that aim to improve robotic and prosthetic devices. One important research field is texture classification based on a neural network inspired by the physiological system of tactile processing. The network structure is composed of primary afferents, a cuneate neuron, and a classification system. In this context, the proposed work aims to evaluate the use of a neuromorphic classifier along with three different networks to classify eight naturalistic textures. The categorization is based on a k-near-est neighbors (KNN) algorithm with a Victor-Purpura distance (VPd) as the distance metric. The results point to the use of a network with slow adapting primary afferents and a KNN with k equal to one as the best system for distinguishing the textures, however, even the worst result was above the 12.5% chance. This system is not only closest to the physiological but also requires less data processing for the classification. We conclude that the VPd-based KNN is a good tool for classification tasks, especially in the case where a more bioinspired and embedded system is required.