University of Dundee
School of Computing
A computer vision system for tracking multiple people in relatively unconstrained environments is described. Tracking is performed at three levels of abstraction: regions, people and groups. A novel, adaptive background subtraction method... more
A computer vision system for tracking multiple people in relatively unconstrained environments is described. Tracking is performed at three levels of abstraction: regions, people and groups. A novel, adaptive background subtraction method that combines colour and gradient information is used to cope with shadows and unreliable colour cues. People are tracked through mutual occlusions as they form groups and part from one another. Strong use is made of colour information to disambiguate occlusions and to provide qualitative estimates of depth ordering and position during occlusion. Some simple interactions with objects can also be detected. The system is tested using indoor and outdoor sequences. It is robust and should provide a useful mechanism for boot-strapping and reinitialisation of tracking using more specific but less robust human models.
Immunoelectron microscopy is used in cell biological research to study the spatial distribution of intracellular macromolecules at the ultrastructural level. Colloidal gold particles (immunogold markers) are commonly used to localise... more
Immunoelectron microscopy is used in cell biological research to study the spatial distribution of intracellular macromolecules at the ultrastructural level. Colloidal gold particles (immunogold markers) are commonly used to localise molecules of interest on ultrathin sections and can be visualised in transmission electron micrographs as dark spots. Quantitative analysis involves detection of the immunogold markers, and is often performed manually
A system providing fall detection and movement monitoring to support older people living at home using computer vision technology is being developed. Sensitive design with user involvement is important if such a system is to be... more
A system providing fall detection and movement monitoring to support older people living at home using computer vision technology is being developed. Sensitive design with user involvement is important if such a system is to be experienced as supportive rather than invasive. Four scenarios, based on material from focus groups and anecdotal evidence, have been developed and performed by a theatre group. These feature older people falling at home with different outcomes and carers discussing an older person's needs. They were filmed and shown to three different groups of older people and a group of sheltered housing wardens to provoke discussion. This method of user requirements gathering provided a shared user context which enabled groups to focus very effectively on the details of a system at the pre-prototyping stages. The results of the discussions are described and the use of this methodology is discussed.
- by Peter Gregor and +2
- •
- Technology, Older people
Drama on video is being used as a tool to investigate user requirements for a fall detector within the context of a monitoring system based on visual tracking. The system is being designed to have the ability to provide passive monitoring... more
Drama on video is being used as a tool to investigate user requirements for a fall detector within the context of a monitoring system based on visual tracking. The system is being designed to have the ability to provide passive monitoring in the homes of older people so that in the case of a fall being detected, the emergency services
Automatic segmentation of bone contours in knee x-ray images is investigated as a step towards reliable, quantitative radiographic analysis of osteoarthritis for diagnosis and assessment of progression. A double contour active shape model... more
Automatic segmentation of bone contours in knee x-ray images is investigated as a step towards reliable, quantitative radiographic analysis of osteoarthritis for diagnosis and assessment of progression. A double contour active shape model is proposed in order to simultaneously segment anterior and posterior contours of the tibial plateaux. Several features are compared for modelling local appearance. Point-to-contour segmentation errors are reported for both femoral and tibial contours.
- by Ian Ricketts and +1
- •
- X-ray imaging, Active Shape Model
Statistical shape models are often learned from examples based on landmark correspondences between annotated examples. A method is proposed for learning such models from contours with inconsistent bifurcations and loops. Automatic... more
Statistical shape models are often learned from examples based on landmark correspondences between annotated examples. A method is proposed for learning such models from contours with inconsistent bifurcations and loops. Automatic segmentation of tibial and femoral contours in knee x-ray images is investigated as a step towards reliable, quantitative radiographic analysis of osteoarthritis for diagnosis and assessment of progression. Results are presented using various features, Mahalanobis distance, distance weighted K−nearest neighbours and two relevance vector machine based methods as quality of fit measure.
A likelihood formulation for detailed human tracking in real world scenes is presented. In this formulation, the appearance, modelled using feature distributions defined over regions on the surface of an articulated 3D model, is estimated... more
A likelihood formulation for detailed human tracking in real world scenes is presented. In this formulation, the appearance, modelled using feature distributions defined over regions on the surface of an articulated 3D model, is estimated and propagated as part of the state. The benefit of such a formulation over currently used techniques is that it provides a dense, highly discriminatory object-based cue that applies in real world scenes. Multi-dimensional histograms are used to represent the feature distributions and an on-line clustering algorithm, driven by prior knowledge of clothing structure, is derived that enhances appearance estimation and computational efficiency. An investigation of the likelihood model shows its profile to be smooth and broad while region grouping is shown to improve localisation and discrimination. These properties of the likelihood model ease pose estimation by allowing coarse, hierarchical sampling and local optimisation.
A method for recovering a part-based description of human pose from single images of people is described. It is able to perform estimation efficiently in the presence significant background clutter, large foreground variation,... more
A method for recovering a part-based description of human pose from single images of people is described. It is able to perform estimation efficiently in the presence significant background clutter, large foreground variation, self-occlusion and occlusion by other objects. This is achieved through two key developments. Firstly, a new formulation is proposed that allows partial configurations, hypotheses with differing numbers of parts, to be made and compared. This permits efficient global sampling in the presence of self and other object occlusions without prior knowledge of body part visibility. Secondly, a highly discriminatory likelihood model is proposed comprising two complementary components. A boundary component improves upon previous appearance distribution divergence methods by incorporating high-level shape and appearance information and hence better discriminates textured, overlapping body parts. An interpart component uses appearance similarity of body parts to reduce the number of false-positive, multipart hypotheses, hence increasing estimation efficiency. Results are presented for challenging images with unknown subject and large variations in subject appearance, scale and pose.
- by Ian Ricketts and +1
- •
- Computer Vision, Prior Knowledge, Pose Estimation, Computer
Breast tissue microarrays (TMAs) facilitate the study of very large numbers of breast tumours in a single histological section, but their scoring by pathologists is time consuming, typically highly quantised, and not without error. This... more
Breast tissue microarrays (TMAs) facilitate the study of very large numbers of breast tumours in a single histological section, but their scoring by pathologists is time consuming, typically highly quantised, and not without error. This paper compares the results of different classification and ordinal regression algorithms trained to predict the scores of immunostained breast TMA spots, based on spot features obtained in previous work by the authors. Despite certain theoretical advantages, Gaussian process ordinal regression failed to achieve any clear performance gain over classification using a multi-layer perceptron. The use of the entropy of the posterior probability distribution over class labels for avoiding uncertain decisions is demonstrated.
A method for automatic segmentation of tumour regions in breast histopathology images is described. It uses auto-context to label pixels based on local image features and contextual label probabilities. We propose spin-context to compute... more
A method for automatic segmentation of tumour regions in breast histopathology images is described. It uses auto-context to label pixels based on local image features and contextual label probabilities. We propose spin-context to compute context features that are invariant under image rotation. Quantitative evaluation is reported using spots stained for estrogen receptor. The use of context resulted in improved segmentation.
- by Telmo Amaral and +1
- •
Tissue microarrays (TMAs) are an important tool in translational research for examining multiple cancers for molecular and protein markers. Automatic immunohistochemical (IHC) scoring of breast TMA images remains a challenging problem. A... more
Tissue microarrays (TMAs) are an important tool in translational research for examining multiple cancers for molecular and protein markers. Automatic immunohistochemical (IHC) scoring of breast TMA images remains a challenging problem. A two-stage approach that involves localization of regions of invasive and in-situ carcinoma followed by ordinal IHC scoring of nuclei in these regions is proposed. The localization stage classifies locations on a grid as tumor or non-tumor based on local image features. These classifications are then refined using an auto-context algorithm called spin-context. Spin-context uses a series of classifiers to integrate image feature information with spatial context information in the form of estimated class probabilities. This is achieved in a rotationally-invariant manner. The second stage estimates ordinal IHC scores in terms of the strength of staining and the proportion of nuclei stained. These estimates take the form of posterior probabilities, enabl...
- by Telmo Amaral and +1
- •
- Pathology Informatics
Breast-tissue microarrays facilitate the survey of very large numbers of tumours but their scoring by pathologists is time consuming, typically highly quantised and not without error. Automated segmentation of cells and intra-cellular... more
Breast-tissue microarrays facilitate the survey of very large numbers of tumours but their scoring by pathologists is time consuming, typically highly quantised and not without error. Automated segmentation of cells and intra-cellular compartments in such data can be problematic for reasons that include cell overlapping, complex tissue structure, debris, and variable appearance. This paper proposes a computationally efficient approach that approximates the density of colour and local invariant features by clusters in the feature space, and characterises each spot by a frequency histogram of nearest cluster centres. Spots are classified into four main types based on their histograms. This approach does not rely on accurate segmentation of individual cells. Classification performance was assessed using 344 spots from the Adjuvant Breast Cancer (ABC) Chemotherapy Trial. A two-layer neural network yielded better classification results than a nearest-neighbour classifier or a single-layer network. Some reasons for classification results in disagreement with pathologist-provided labels are discussed and include the existence of spots containing large proportions of different tissue types.
- by Telmo Amaral and +1
- •
- Breast Cancer
Tissue microarrays have become an important tool in clinical research to analyse molecular and protein markers in various types of cancer. However their analysis is a timeconsuming task and introduces inter-and intra-observer variations.... more
Tissue microarrays have become an important tool in clinical research to analyse molecular and protein markers in various types of cancer. However their analysis is a timeconsuming task and introduces inter-and intra-observer variations. An automated method is proposed, called spin-context, to segment in-situ and invasive tumour regions in images of breast tissue microarrays. Spin-context incorporates contextual information extracted from images in a rotationally invariant manner. Additionally, the effect of removing background context locations at boundaries of tissue microarray spots is evaluated. Quantitative evaluation is reported using tissue microarray spots stained for estrogen receptor. Results show that incorporating context in this way improves classification performance, particularly around spot boundaries, compared to classification incorporating no context.
- by Telmo Amaral and +1
- •
Breast tissue microarrays (TMAs) facilitate the study of very large numbers of breast tumours in a single histological section, but their scoring by pathologists is time consuming, typically highly quantised, and not without error. This... more
Breast tissue microarrays (TMAs) facilitate the study of very large numbers of breast tumours in a single histological section, but their scoring by pathologists is time consuming, typically highly quantised, and not without error. This paper compares the results of different classification and ordinal regression algorithms trained to predict the scores of immunostained breast TMA spots, based on spot features obtained in previous work by the authors. Despite certain theoretical advantages, Gaussian process ordinal regression failed to achieve any clear performance gain over classification using a multi-layer perceptron. The use of the entropy of the posterior probability distribution over class labels for avoiding uncertain decisions is demonstrated.
Root growth in the field is often slowed by a combination of soil physical stresses, including mechanical impedance, water stress, and oxygen deficiency. The stresses operating may vary continually, depending on the location of the root... more
Root growth in the field is often slowed by a combination of soil physical stresses, including mechanical impedance, water stress, and oxygen deficiency. The stresses operating may vary continually, depending on the location of the root in the soil profile, the prevailing soil water conditions, and the degree to which the soil has been compacted. The dynamics of root growth responses are considered in this paper, together with the cellular responses that underlie them. Certain root responses facilitate elongation in hard soil, for example, increased sloughing of border cells and exudation from the root cap decreases friction; and thickening of the root relieves stress in front of the root apex and decreases buckling. Whole root systems may also grow preferentially in loose versus dense soil, but this response depends on genotype and the spatial arrangement of loose and compact soil with respect to the main root axes. Decreased root elongation is often accompanied by a decrease in both cell flux and axial cell extension, and recent computer-based models are increasing our understanding of these processes. In the case of mechanical impedance, large changes in cell shape occur, giving rise to shorter fatter cells. There is still uncertainty about many aspects of this response, including the changes in cell walls that control axial versus radial extension, and the degree to which the epidermis, cortex, and stele control root elongation. Optical flow techniques enable tracking of root surfaces with time to yield estimates of twodimensional velocity fields. It is demonstrated that these techniques can be applied successfully to timelapse sequences of confocal microscope images of living roots, in order to determine velocity fields and strain rates of groups of cells. In combination with new molecular approaches this provides a promising way of investigating and modelling the mechanisms controlling growth perturbations in response to environmental stresses.
We present a method for learning appearance models that can be used to recognise and track both 3D head pose and identities of novel subjects with continuous head movement across the view-sphere. We describe an automatic face data... more
We present a method for learning appearance models that can be used to recognise and track both 3D head pose and identities of novel subjects with continuous head movement across the view-sphere. We describe an automatic face data acquisition system based on a magnetic sensor and a calibrated camera. The system enabled us to obtain systematically a database of face images with labelled 3D poses across a view-sphere of £ ¥ ¤ § ¦ § yaw and £ ¥ © ¦ tilt at intervals of ¦. The database was used to learn appearance models of unseen faces based on similarity measures to prototype faces. The method is computationally efficient and enables real-time performance.