Dr. D. Dubovitskiy the company director of Oxford Recognition Ltd is an expert in artificial intelligence, machine vision and pattern recognition. His work has involved hardware and software engineering, system design, mathematical modeling, image processing, digital filtering, and digital signal processing algorithm construction. The company has developed AI systems comprising data acquisition modules, ADSP, embedded software, PC controllers, system drivers, graphical user interfaces and data storage systems. The company has considerable experience in project management, technology transfer and in the practical implementation of scientific and technological results.
The company has fourteen years of experience in Digital Image Processing. Since 2000 we have been designing and developing software and hardware solutions for a variety of different applications. We have created surface inspection systems for an iron and steel corporation, an early cancer detection system for medical imaging, image enhancement algorithms, a digital watermarking process for audio and video information and a security application with independent protected core functioning.
For our first project we designed a defect recognition and analysis system for an iron and steel corporation. This project used line scan cameras for scanning moving metal strips which were moving at a speed of up to 6 m/sec. The detector hardware then highlighted regions of interest and the analytic software made a decision about particular object.
Our second research and development project involved the design of a medical imaging system in order to screen and recognize abnormal biological cells in Cervical Smear test samples. The system used microscopy with a motorized table. The sample was prepared by the Papanicolau method and then automatically scanned by a microscope which was capable of making a decision on whether the disease was present.
Another similar application of the Cervical Smear diagnosis algorithm was in skin cancer detection. This project focused on the use of the algorithm by GP doctors for evaluation of suspicious birth-marks or moles found on patients. The image acquisition was performed with a Nikon dermatology lens adapter and the structure recognition descriptors were designed in this case to push the decision making engine to make an optimal-probabilistic conclusion for particular skin conditions.
Most recently we have been working on several small projects involving image enhancement systems. The most significant of these is the Correspondence point matching, Stereo reconstruction procedure project in which an eye tracking device is currently under development thanks to financial support from the EU.
Our third major project has produced significant developments in the areas of hardware - a USB Mass Storage device designed for security applications will allow users to have independent control of data which has been saved on it.
The forth active division involves specially designed Digital Watermarking algorithms for adding digital seal information or owner specific data to pictures or audio files. This approach allows users to add small quantities of extra data to a media file. The most interesting feature of the given solution is that information can be recovered even after reasonable compression or passing through a noise channel.
The company is actively engaged in applying the image processing techniques we have developed in the Image processing field. In addition we have our own technology in digital Fractal Geometry and a Fuzzy Logic system for autonomous object recognition activity. It is involved in special learning procedures in the structural analysis and navigation field. Several 2D binary original operators have been discovered and the intellectual property rights have been published.