There are many approaches dealing with various aspects of texture analysis, segmentation and classification. One of the characteristics among most of these approaches and those using neighborhood is that they require applying a template to a given image, pixel by pixel, to yield a new image. Although the selection of an appropriate window size is critical and affects directly the results obtained, it is often done arbitrarily. We present, in this chapter, a new approach based on the concept of variography to achieve the selection of the optimal window. We develop some direct applications including textural primitives description, mathematical morphology, textural segmentation and textural classification. Promising results are obtained and presented.