Nick Kingsbury‘s research spans the following areas:
Speech compression (using multi-pulse and code-excited linear prediction models);
Error resilient vector quantisation techniques for speech compression;
Combined coding and modulation for FM data transmission;
Optimal data demodulation methods for channels with phase jitter;
Image and video sequence compression using sub-band filter banks, lapped transforms, wavelets and lattice vector quantisation;
Error resilient techniques for image and video compression, to provide graceful degradation over noisy channels and avoid the need for error correction coding where possible;
Human visual system models for measuring visibility of coding artefacts;
Motion estimation techniques using complex lapped transforms and wavelets;
Motion estimation methods which correctly model occlusion and relative motion of multiple objects;
3-D object data acquisition methods for application to face recognition.
Complex Wavelet Transform techniques.
Applications of complex wavelets to denoising, texture analysis / synthesis, image classification and segmentation, and compression.
Robust watermarking of images and video in the complex wavelet domain.
Denoising and segmentation of 3-D datasets using complex wavelets.
Seismic image reconstruction.
Content-based image retrieval.
Magnetic resonance and ultrasound image enhancement and visualisation.
Registration of non-rigid objects in images and 3D data.
Robust keypoint detection and local feature descriptors for object recognition.