Transforming images to data


A lot of today’s technological advances are based on the ability to transfer data between different formats, such as images, written text, audio and stored data. The increase in digital media has assisted in collecting this information more readily but the transfer of data from images or audio systems are still problematic. Visual analysis of data can pose specific issues if the identification of features and zones are required as image quality and orientation can change between subsequent images. At GeoSquare we were posed with an interesting problem to extract information from a number of images that contained deep sea nodules lying on the sea floor, with our experience in image processing we looked at numerous options to resolve the issue.

Nodules on the Seabed (Abramax CC3)

Nodules on the Seabed (Abramax CC3)

Identification of nodules on sea floor

Identified Deep Sea Nodules on the Seabed

Visual analysis of data requires that the image is processed using an algorithm that could deal with different variations in each image while maintaining quality control on data collected. Our system affords high throughput analysis without the requirement of supercomputing facilities and can be performed in different work environments.

Data obtained from the visual analysis system can be prepared in a number of suitable formats for use by professionals to perform further analysis. Output associated with the visual analysis of data includes parameters such as coverage, major axis, orientation and location in addition to images with overlaid data, histograms and cluster analysis.

The key benefits of using this technology with deep sea nodules are optimized mining target areas which could be extended to grade control and the ability to minimize the impact on the sub-sea environment. Work is ongoing to improve this assessment methodology but our first set of analysis has recently been published (link) in conjunction with a resource assessment.