The document proposes an anomaly detection scheme for hyperspectral images based on a non-Gaussian mixture model using a Student's t-distribution. It estimates the background probability density function using a Bayesian approach that models each pixel as a mixture of Student's t distributions. The anomaly detection strategy then applies a generalized likelihood ratio test. Experimental results on real hyperspectral data show the proposed Bayesian Student's t mixture model can reliably estimate the background distribution and effectively detect anomalous objects, outperforming a Gaussian mixture model approach.