Pixel-based correspondence for facial images

A pixel-based correspondence method is presented for representation of facial images. The proposed method consists of two modules: face-image matching and face-image morphing. In the face-image matching module, the correspondence between two images are calculated for all pixel locations. A novel area-based matching method is proposed that makes use of the concept of the fractal dimension, and develops a non-parametric local transform as a basis for establishing correspondence between two face images. In the face-image morphing module, a mapping is performed for deformation of the source face image onto the target face image. This is done to map the pixels in the source face image to the location of their corresponding pixels in the target image.

Fuzzy measures of pixel cluster compactness

Pixel-scale fine details are often lost during image processing tasks such as image reduction and filtering. Block or region based algorithms typically rely on averaging functions to implement the required operation and traditional function choices struggle to preserve small, spatially cohesive clusters of pixels which may be corrupted by noise. This article proposes the construction of fuzzy measures of cluster compactness to account for the spatial organisation of pixels. We present two construction methods (minimum spannning trees and fuzzy measure decomposition) to generate measures with specific properties: monotonicity with respect to cluster size; invariance with respect to translation, reflection and rotation; and, discrimination between pixel sets of fixed cardinality with different spatial arrangements. We apply these measures within a non-monotonic mode-like averaging function used for image reduction and we show that this new function preserves pixel-scale structures better than existing monotonie averages.

A pixel-based approach to estimation of solar energy potential on building roofs

Precise estimation of solar energy on building roofs plays a critical role in sustainable development and renewable energy consumption of high-density human habitats. Conventional solar radiation models based on costly Light Detection and Ranging (LiDAR) data are only adequate for existing buildings, not for future construction areas. In this paper, a pixel-based methodology is constructed for estimating solar energy potential over roofs. Buildings with flat roofs in a newly planned construction area are chosen as a case study. The solar radiation at a certain cell is mathematically formulated in the pixel unit, and its yields over a certain time period are calculated by considering multiple instantaneous solar irradiances and are visually presented by image processing. Significant spatial and temporal variations in solar radiation are measured. Within the study area, the maximum and minimum annual radiation yields are estimated at 4717.72 MJ/m2/year and 342.58 MJ/m2/year respectively. Radiation contour lines are then mapped for outlining installation ranges of various solar devices. For each apartment building, around 20% of roof areas can obtain 4500 MJ/m2/year or more solar radiation yields. This study will benefit energy investors and urban planners in accurately predicting solar radiation potential and identifying regions with high radiation over building roofs. The results can be utilised in government policies and urban planning to raise awareness of the use of renewable energy sources.

Pixel review: Dance and digital wizardry conjure stunning illusions

A number of artists have explored the fusion of technology with performing arts. From Gideon Obarzanek​ and Frieder Weiss' projection mapping techniques with Glow and Mortal Engine to Garry Stewart and Thomas Pachoud's​ experiments with real-time looping and layered video in Proximity, it seems that infusing performance with choreographed bodies, screen-based design and choreographed pixels is the next frontier of contemporary performance.

Multiresolution analysis of connectivity

Multiresolution histograms have been used for indexing and retrieval of images. Multiresolution histograms used traditionally are 2d-histograms which encode pixel intensities. Earlier we proposed a method for decomposing images by connectivity. In this paper, we propose to encode centroidal distances of an image in multiresolution histograms; the image is decomposed a priori, by connectivity. Multiresolution histograms thus obtained are 3d-histograms which encode connectivity and centroidal distances. The statistical technique of Principal Component Analysis is applied to multiresolution 3d-histograms and the resulting data is used to index images. Distance between two images is computed as the L2-difference of their principal components. Experiments are performed on Item S8 within the MPEG-7 image dataset. We also analyse the effect of pixel intensity thresholding on multiresolution images.

Comparison of MLP and Bayesian approaches on mineral prospectivity mapping tasks

Mineral Prospectivity Mapping is the process of combining maps containing different geoscientific data sets to produce a single map depicting areas ranked according to their potential to host mineral deposits of a particular type. This paper outlines two approaches for deriving a function which can be used to assign to each cell in the study area a value representing the posterior probability that the cell contains a deposit of the sought-after mineral. One approach is based on estimating probability density functions (pdfs); the second uses multilayer perceptrons (MLPs). Results are provided from applying these approaches to geoscientific datasets covering a region in North Western Victoria, Australia. The results demonstrate that while both the Bayesian approach and the MLP approach yield similar results when the number of input dimensions is small, the Bayesian approach rapidly becomes unstable as the number of input dimensions increases, with the resulting maps displaying high sensitivity to the number of mixtures used to model the distributions. However, despite the fact that Bayesian assigned values cannot be interpreted as posterior probabilities in high dimensional input spaces, the pixel favorability rankings produced by the two methods is similar.

Constructing artificial images of facial expressions

This paper presents a method for construction of artificial images of facial expressions. The proposed fractal-based synthesis procedure called pixel-based correspondence works on 2D images and does not require any depth information. This method can generate artificial images of an object when only a single image is given. Using the proposed method, effective example-based facial analysis systems can be trained and utilised in various applications.

Simulating autonomous robot teams with Microsoft robotics studio

This paper presents an application of Microsoft Robotics Studio (MSRS) in which a team of six four wheel drive, ground based robots explore and map simulated terrain. The user has the ability to modify the terrain and assign destination objectives to the team while the simulation is running. The terrain is initially generated using a gray scale image, in which the intensity of each pixel in the image gives an altitude datum. The robots start with no knowledge of their surroundings, and map the terrain as they attempt to reach user-defined target objectives. The mapping process simulates the use of common sensory hardware to determine datum points, including provision for field of view, detection range, and measurement accuracy. If traversal of a mapped area is indicated by the users’ targeting commands, path planning heuristics developed for MSRS by the author in earlier work are used to determine an efficient series of waypoints to reach the objective. Mutability of terrain is also explored- the user is able to modify the terrain without stopping the simulation. This forces the robots to adapt to changing environmental conditions, and permits analysis of the robustness of mapping algorithms used when faced with a changing world.

A lightwave 3d plug-in for modelling long hair on virtual humans

Multimedia applications today make use of virtual humans. Generating realistic virtual humans is a challenging problem owing to a number of factors, one being the simulation of realistic hair. The difficulty in simulating hair is due to the physical properties of hair. The average human head holds thousands of hairs, with the width of each hair often smaller than the size of a pixel. There are also complex lighting effects that occur within hair. This paper presents a LightWave 3D plug-in for modeling thousands of individual hairs on virtual humans. The plug-in allows the user to specify the length, thickness and distribution of the hair, as well as the number of segments a hair is made up of. The plug-in is able to add hairs to a head model, which the user then modifies to define a hairstyle. The hairs are then multiplied by the plug-in to produce many hairs. By providing a plug-in that does most of the work and produces realistic results, the user is able to produce a hairstyle without modeling each individual strand of hair. This greatly reduces the time spent on hair modeling, and makes the possibility of adding realistic long hair to virtual humans reasonable.

Implementation and applications of the distortion operator

The distortion operator transforms 2D images in a manner similar to image warping or morphing, allowing source pixels to be mapped to any destination pixel. This operator can be implemented on current hardware, allowing at least one distortion per frame at interactive frame rates. Potential applications are numerous, but those described include re-mapping images for correct projection onto curved screens, correcting camera distortion from multiple sources simultaneously, and allowing constant time dynamic texturing and lighting of a static scene which is independent of geometric complexity.

Image magnification using interval information

In this paper, a simple and effective image-magnification algorithm based on intervals is proposed. A low-resolution image is magnified to form a high-resolution image using a block-expanding method. Our proposed method associates each pixel with an interval obtained by a weighted aggregation of the pixels in its neighborhood. From the interval and with a linear K_α operator, we obtain the magnified image. Experimental results show that our algorithm provides a magnified image with better quality (peak signal-to-noise ratio) than several existing methods.

Representing images by means of interval-valued fuzzy sets. Application to stereo matching

Stereo matching tries to find correspondences between locations in a pair of displaced images of the same scene in order to extract the underlying depth information. Pixel correspondence estimation suffers from occlusions, noise or bias. In this work, we introduce a novel approach to represent images by means of interval-valued fuzzy sets to overcome the uncertainty due to the above mentioned problems. Our aim is to take advantage of this representation in the stereo matching algorithm. The image interval-valued fuzzification process that we propose is based on image segmentation in a different way to the common use of segmentation in stereo vision. We introduce interval-valued fuzzy similarities to compare windows whose pixels are represented by intervals. In the experimental analysis we show the goodness of this representation in the stereo matching problem. The new representation together with the new similarity measure that we introduce shows a better overall behavior with respect to other very well-known methods.

Color image magnification with interval-valued fuzzy sets

In this work we present a simple magnification algorithm for color images. It uses Interval-Valued Fuzzy Sets in such a way that every pixel has an interval membership constructed from its original intensity and its neighbourhood's one. Based on that interval membership, a block is created for each pixel, so this is a block expansion method.

Model-based combined tracking and resolution enhancement

Wide area surveillance requires high-resolution images of the object of interest derived possibly from only low-resolution video of the whole scene. We propose a combined tracking and resolution enhancement approach that increases the resolution of the object of interest during tracking. The key idea is the use of an object-specific 3D mesh model with which we are able to track non-planar objects across a large number of frames. This model is subdivided such that every triangle is smaller than a pixel when projected into the image to facilitate super-resolution on the model rather than on the image. We apply our approach to faces and show that it outperforms interpolation methods by achieving resolution enhancement, while being less blurred.