The development of machine algorithms for the Cranfield underwater laser stripe imaging system

When using conventional illumination in the underwater environment, the greatest problem faced has always been that of back-scatter. This reduces visual quality both to the naked eye and through the use of underwater cameras. This project continues research into a hybrid underwater laser viewing system, combining a laser scanning technique with conventional underwater viewing system technology. The use of carefully positioned lighting can greatly reduce the problem of back-scattered light, improving image definition and contrast. With the use of a laser light source this can be improved even further within a more flexible system. Utilising two scanning devices a narrow stripe of light can be formed which is scanned over a target area. Small successive regions of any object lying within that area can then be illuminated, isolated and extracted using image processing to compile a complete image of the object. For the purposes of surf zone and shallow water mines counter measures, the system requires reliable machine algorithm based image processing techniques, to compile a constructed image of target objects within highly turbid regions. Projected stripes take on various characteristics dependent on the geometry and aspect of the object which they fall upon and the conditions experienced. High levels of turbidity leading to significant noise, creates considerable difficulties in the isolation of the stripe region and the extraction of meaningful visual information. The success of any technique would rely upon the overcoming of these difficulties, allowing the system to offer the operator the most useful visual information permissible in the final compilation of the image. Through various image processing techniques these factors have been tackled and their effects on the final image, greatly reduced. The result is the creation of highly stable and effective processing techniques allowing improved object definition and recognition, in levels of high turbidity.