Gaussian mixture modeling is a recent approach in texture analysis and is used to model image textures. Texture is modeled using a mixture of Gaussian distributions, which capture the local statistical properties of the texture. The mixture parameters are estimated using Expectation Maximization algorithm. This algorithm finds the maximum likelihood estimate of the parameters of an underlying distribution from a given data set when data is incomplete. The paper presents a method of identifying changes as well as new patterns in the image using the Gaussian mixture model parameters. Model parameters of the original image texture are computed. Unexpected patterns in the image are discriminated by using weighted normalized Euclidean distance measure derived from the model parameters.