Gerhard de Jager

Shape-from-Silhouette with two mirrors and an uncalibrated camera

Two planar mirrors are positioned to show five views of an object, and snapshots are captured from different viewpoints. We present closed form solutions for calculating the focal length, principal point, mirror and camera poses directly from the silhouette outlines of the object and its reflections. In the noisy case, these equations are used to form initial parameter estimates that are refined using iterative minimisation. The self-calibration allows the visual cones from each silhouette to be specified in a common reference frame so that the visual hull can be constructed. The proposed setup provides a simple method for creating 3D multimedia content that does not rely on specialised equipment. Experimental results demonstrate the reconstruction of a toy horse and a locust from real images. Synthetic images are used to quantify the sensitivity of the self-calibration to quantisation noise. In terms of the silhouette calibration ratio, degradation in silhouette quality has a greater effect on silhouette set consistency than computed calibration parameters.

Automatic registration of temporal image pairs for digital subtraction angiography

Temporal Digital Subtraction Angiography (DSA) is used to visualize blood vessels in x-ray images. A DSA image pair consists of the mask image, which is a digitized x-ray taken before a contrast medium is injected into the bloodstream, and the live image, which is taken once the contrast medium has traversed the circulatory system and reached the blood vessels of interest. The mask image is then subtracted from the live image and ideally only the contrast enhanced blood vessels should remain. DSA has two main limitations. Firstly, gross patient motion and physiological events occur in the time that elapses between x-rays. Secondly, there are local and global differences in the mean gray-level at corresponding points in the live and mask images, excluding the variations introduced by the contrast media. To solve the motion problem, we take the approach of matching regions around control points in the live image in a search area around the approximately corresponding points in the mask image. In this way a motion vector field that describes the spatial offset to the best match position in the mask image (with subpixel accuracy) is constructed. The problem of mean gray- level disparity between the live and mask images is to a large extent overcome by the use of a match measure that is invariant to overall additive gray-level differences. Incorrect mismatches caused by the contrast media are avoided by using multiple subtemplates in the matching process. The subtemplate method also allows the estimation of mean gray-level disparity between the mask and live images. The smoothed motion vector field and mean gray-level disparity estimates are used to perform an improved subtraction of the mask from the live image with a reduction in the artifacts that are a result of normal subtraction. Efficient best match search techniques are used to reduce the computational cost of the algorithm, at the expense of some difference image quality. Results are provided for simulated and actual DSA image pairs.

An explanation for the extremely low, but variable, radiation dosages measured in a linear slit scanning radiography system

Clinical trials performed for the FDA’s Section 510k compliance submission of the Statscan digital, full-body, linear slit scanning diagnostic radiography system revealed that comparable diagnostic results with a commercial full-field screen film device were obtained with the Statscan using much lower radiation doses. For certain imaging procedures the doses for Statscan were as much as twenty to thirty times lower. However the results varied by a large amount and in particular the results for chest radiographs were anomalous in that the Statscan dose was less reduced. Whilst it is well known that slit scanning radiography has considerably lower radiation exposure than full-field devices due to its much lower scatter to primary ratio and also that digital radiography has the potential for lower radiation dosages, it was thought that that this alone did not fully account for the dose differences. This paper suggests that these dose differences, including the anomaly mentioned above, can be explained by considering the unique way that slit scanning is undertaken by Statscan i.e. by scanning the tube, detector, slit and collimators together along a linear path. The effect on measured skin entrance doses is explained and the dosage differences as affected by digital technology, higher DQE, slit scanning (low scatter to primary ratio) and linear slit scanning methods are quantified. Furthermore it is explained how the Statscan geometry leads an improved “skin sparing” effect.

Optimality in detecting targets with unknown location

An optimal test does not exist for the problem of detecting a known target with unknown location in additive Gaussian noise. A common solution uses a generalised likelihood ratio testing (GLRT) formalism, where a maximum likelihood estimate of the unknown location parameter is used in a likelihood ratio test. The performance of this test is commonly assessed by comparing it to the ideal matched filter, which assumes the target location known in advance. This comparison is of limited utility, however, since the fact that the location is unknown has a significant effect on the detectability of the target. We demonstrate that a uniformly most powerful invariant (UMPI) optimal test exists for a specific class of unknown target location problems, where observations are discrete and shifts are defined circularly. Since this approach explicitly models the location as unknown, an assessment of the suboptimality of competing tests becomes meaningful. It is shown that for certain examples in this class the GLRT performance is negligibly different from that of the optimal test.

Detective quantum efficiency of the LODOX system

The Detective Quantum Efficiency (DQE) of a digital x-ray imaging system describes how much of the signal to noise ratio of the incident radiation is sustained in the resultant digital image. This measure of dose efficiency is suitable for the comparison of detectors produced by different manufacturers. The International Electrotechnical Commission (IEC) stipulates standard methods and conditions for the measurement of the DQE for single exposure imaging systems such as flat panel detectors. This paper shows how the calculation is adapted for DQE measurements of scanning systems. In this paper it is described how to measure the presampled Modulation Transfer Function (MTF) using an edge test method and how to extract the horizontal and vertical components of the Noise Power Spectrum (NPS) in a way that is insensitive to structured noise patterns often found in scanned images. The calculation of the total number of incident photons from the radiation dose measurement is explained and results are provided for the Lodox low dose full body digital x-ray scanning system which is developed in South Africa.The Detective Quantum EÆciency (DQE) of a digital x-ray imaging system describes how much of the signal to noise ratio of the incident radiation is sustained in the resultant digital image. This measure of dose eÆciency is suitable for the comparison of detectors produced by di erent manufacturers. The International Electrotechnical Commission (IEC) stipulates standard methods and conditions for the measurement of the DQE for single exposure imaging systems such as at panel detectors. This paper shows how the calculation is adapted for DQE measurements of scanning systems. In this paper it is described how to measure the presampled Modulation Transfer Function (MTF) using an edge test method and how to extract the horizontal and vertical components of the Noise Power Spectrum (NPS) in a way that is insensitive to structured noise patterns often found in scanned images. The calculation of the total number of incident photons from the radiation dose measurement is explained and results are provided for the Lodox low dose full body digital x-ray scanning system which is developed in South Africa.

Model-based segmentation of medical x-ray images

This paper discusses the methods used to model the structure of x-ray images of the human body and the individual organs within the body. A generic model of a region is built up from x-ray images to aid in automatic segmentation. By using the ribs from a chest x-ray image as an example, it is shown how models of the different organs can be generated. The generic model of the chest region is built up by using a priori knowledge of the physical structure of the human body. The models of the individual organs are built up by using knowledge of the structure of the organs as well as other information contained within each image. Each image is unique and therefore information from the region surrounding the organs in the image has to be taken into account when adapting the generic model to individual images. Results showing the application of these techniques to x-ray images of the chest region, the labelling of individual organs, and the generation of models of the ribs are presented.

SIMD implementation of integer DCT and hadamard transforms in H.264/AVC encoder

This paper investigates Single Input Multiple Data (SIMD) implementations of the 4×4 integer Discrete Cosine Transforms (DCT) and 4×4 Hadamard transforms in H.264/AVC video encoders. Integer DCT transforms decomposes the video signal from the spatial or image domain to the frequency domain while the Hadarmad transforms decomposes the luminance and chrominance component of the same video signal. Results show that for the same Peak Signal to Noise Ratio (PSNR) and at quality factor of eight (8), integer DCT and Hadamard transforms have a computation time gain of 14000ms and 17000ms representing a speed up of 45% and 50% compared to the matrix multiplication method respectively. At these computation time gains our implementation is suited for use in real-time video broadcasting/internet applications and in low bandwidth environments.

Real-time machine vision inspection of plastic bottle closures

This paper presents a successful implementation of a real time inspection system of plastic bottle closures. The closures are inspected at a rate of 20 per second or one every 50ms. The available time to inspect each closure forces the algorithms used to be relatively simple. Even though the algorithms are simple, they have to be robust to ensure a good result. Two boundary tracking algorithms were designed and implemented, one based on edge strength information and one based on threshold information. Both use prior knowledge about the closure. The results are sufficiently accurate to replace a human operator with the machine inspection system. A description of the system including the timing and hardware used is given. The results achieved using each of the algorithms are presented. This is followed by a brief discussion of problems associated with achieving highly accurate results using a simple algorithms.

A Linear Image-Pair Model and the Associated Hypothesis Test for Matching

A statistical model is developed for the image pair and used to derive a minimum-error hypothesis test for matching. For reasons of tractability a multivariate normal image model and linear dependence between images are assumed. As one would expect, the optimal test out-performs the standard approaches when the assumed model is in force, but the extent of the optimal tests superiority suggests that there is significant potential for improvement on the standard methods of assessing image similarity.

Mesh IP Video Surveillance Systems Model Design and Performance Evaluation

This paper proposes a model for Internet Protocol (IP) mesh video surveillance systems and presents the performance evaluation of Ah-hoc On-demand Distance Vector (AODV) and Open Shortest Path First (OSPF) routing protocols in wireless mesh video surveillance system. A wireless mesh video surveillance network consists of IP cameras linked to mesh routers which are then linked to the mesh gateways. Local monitoring can be done by inserting a switch or router between the gateways and the Internet while remote monitoring can be done through the Internet. Routing provides selection, constructing and management of routes in order to maximize throughput and minimize video packet losses, end-to-end delays and Jitter. Results show that the OSPF routing protocol outperformed the AODV in throughput, packet loss, end-to- end delay and Jitter terms with a throughput advantage of 35%.