Spatial Association Between Images

Digital images are subject to a variety of contaminations (distortions) during the acquisition, processing, compression, storage transmission, and reproduction. This can significantly affect the posterior visualization of images. In image processing there are at least two ways to approach this issue, objective and subjective image quality assessment. Several authors have stressed the inconvenients of working with subjective evaluation (see, for instance, Wang et al. 2004). In the objective quality assessment approach, the goal is to develop measures that can quantify the image quality, most of the time through a similarity index that takes into account different aspects of the images such as texture, color, contrast among others. One of the most popular reference quality indices is the mean square error (MSE), which is computed averaging the square of the intensity differences pixel to pixel such that large differences are associated with departures from similarity or agreement between the images. In the same spirit the peak signal-to-noise ratio (PSNR) index quantifies the ratio between the maximum possible power of a signal and the power of corrupting noise. Two important features of these measures of similarity is that they are simple to calculate and have a physical meaning. One disadvantage of the MSE and PSNR coefficients, if the fact that none of them is able to capture the human visual system (HVS) (Eskicioglu and Fisher 1995). Some proposals in this context have been designed to modify the MSE penalizing the errors in accordance with their visibility.