No. K-nearest neighbor is a supervised approach where each example to be classified is put into a class by looking at the classes of the K nearest training examples and assigning the example to the majority class. When you say nearest centroids I believe you are referring to methods like K-means or K-medioids? These are unsupervised clustering methods where K centroids are initialized into your space and the examples to be classified are put into the class with the nearest centroid, then the centroids are re-assigned, and the process is repeated iteratively until convergence.