Dr. Timor Kadir

Robotics Research Group
Dept of Engineering Science
Ewert House, Ewert Place.
Summertown, Oxford
OX2 7DD

Tel : (01865) 280943
Fax : (01865) 273908

Timor Kadir was a research assistant in the Robotics Research Group in Oxford University. He is now a research scientist at Siemens Molecular Imaging (formally Mirada Solutions) in Oxford. He still reads his university email and occasionally maintains this website.

Research Interests

His DPhil thesis was entitled Scale, Saliency and Scene Description and was concerned with the Scene Description task --- the automatic extraction of a set of robust, relevant, and su fficiently complete semantic descriptions of a scene, for subsequent inference. To this end, a novel algorithm called Scale Saliency, for quantifying image region saliency was presented. In this new approach, regions are considered salient if they are simultaneously unpredictable both in some feature and scale-space. Unpredictability is determined as a function of the local PDF, generating a space of saliency values over x, y and scale, fro m which features may be extracted by a suitable detection strategy. The technique is a more generic approach to scale and saliency compared to conventional methods, because both are defined independent of a particular basis morphology. The method can be made invariant to rotation, translation, non-uniform scaling, and uniform intensity variations and robust to small changes in viewpoint.

The algorithm was applied to simple recognition tasks and the features shown to be robust and persistent (hence useful for tracking). The relevance of the scales and the generality of saliency demonstrated by using the PDF of salient scales to cha racterise textures in a technique called Scale Descriptors. For the texture segmentation experiments, a novel unsupervised Level Set based implementation of Region Competition was developed. The key aspect of this is that it operates on just one surface and uses a non-parametric region model.

Finally, a unified approach to image modelling was proposed based on two scales of spatial (un)predictability --- the local and semi-local. Quantifying the (un)predictability of image attributes at these two scales enables a space of image models that can represent several different image content types such as blobs, lines, statistical and structural textures in a unified framework.

Here you can find more information on:

The bit of Robotics that he is part of used to be in the Terrapin Hut. Sadly, this is no longer the case and the Terrapin is now housing some builders, and all the former Terrapanoids are in some squashed new temporary lab with that MVL lot.

By the way, if you are installing Matlab on a PC running Linux, then you might need this fix : FixMatlabMex.