University of Ulster: A Novel Spiral Addressing Scheme for Rectangular Images
As part of the research for Slándáil, Ulster University have been working on a novel method for analysing images based on a system that they have developed. The goal of this is to make image analysis from social media more efficient, taking into account the large volume of images that are shared during a natural disaster (for example, Instagram reported 1.3 million pictures posted during Hurricane Sandy).
Communication via accurate, complete and real-time information sharing is key to prepare, respond and recover in disaster management. Sharing visual content not only increases the credibility of the information, but also encourages social media user engagement.
For most existing web search platforms, such as Bing, Google, and Yahoo, search is based on context information, i.e., tags, time or location. Text-based search is fast and convenient, but the search results can be mismatched, less relevant, or duplicated due to web noise. Therefore, incorporating content-based analysis, such as image analytics, can improve the search quality.
To help to achieve the balance of speed and accuracy for large-scale visual information retrieval, the research conducted by University of Ulster aims to develop a novel framework to facilitate real-time processing for web image and video analysis based on characteristics of the human vision system, and designed originally for identifying events that are indicative of emerging security threats.
The Proposed Method
In Ulster’s recent work, a novel spiral image processing framework has been proposed that develops an efficient spiral addressing scheme for standard square images. We refer to this new framework as “squiral” (square spiral) image processing (SIP).
Spiral architectures have been employed as an efficient addressing scheme in hexagonal image processing (HIP), whereby the image pixel indices can be stored in a one-dimensional vector that enables fast image processing. However, this computational advance of HIP is hindered by the additional time and effort required for conversion of image data to a HIP environment as existing hardware for image capture and display are based predominantly on traditional rectangular pixels.
Unlike HIP, conversion to the SIP addressing scheme can be achieved easily using an existing lattice with a Cartesian coordinate system; there is also no need to design special image processing operators. Furthermore, the researchers have developed a SIP-based non-overlapping convolution technique by simulating the “eye tremor” phenomenon of the human visual system, which facilitates fast computation. The proposed SIP-based framework is shown below.
For illustration above, this technique has been implemented for the purpose of edge detection. The preliminary results demonstrate the efficiency of the SIP framework by comparison with standard 2D convolution and separable 2D convolution.