A diverse cross section of the society uses digital image and video data ranging from entertainment to medicine, from education to surveillance. In the last decade the number of image and video based applications servicing these industries has increased by many folds. Consequently the amount of image and video data that needs to be processed is rising exponentially. Hence, there is a constant quest for techniques and methodologies that could assist the users of these applications in efficiently analysing the vast amount of image and video data. Researchers have reduced the problem of image and video analysis into smaller sub-problems and a lot of attention has been given to detection of fundamental semantic building blocks of images or videos. This thesis delves on an omnipresent semantic building block, that is texture. Even though humans can recognise textures, they find it quite difficult to verbalise it. It is also quite impossible to formulate a universal mathematical model for describing textures. However, there are some desirable invariant properties that a texture descriptor should possess to transformations such as scale, illumination and rotation etc. We propose a solution to the problem of local-texture description by deriving properties using both, statistics and structure-based approaches whilst making sure that the descriptors are invariant to transformations such as rotation, illumination, etc. Point-based image matching is one of the techniques used in image-analysis, we propose a framework for incorporating these local-texture descriptors into an image-matching framework. We also realise a moving-object detector, which is one of the fundamental processes used in video analysis. As for moving-object detection, some of the invariant properties of the local-texture descriptor are redundant, hence we modify the proposed local-texture descriptor to incorporate only the requisite invariant properties. Further more, we present innovative frameworks for a real-world application using these local-texture descriptors and other fundamental processes being proposed in this thesis.