Jose Luis Silvan-Cardenas

Assessing fine-spatial-resolution remote sensing for small-area population estimation

Small-area population estimation is an important task that has received considerable attention from the remote-sensing community in the past four decades. The wealth of related studies reveals that the notion of living space had been considered a key linkage between population and remote-sensing measurements. Unfortunately, a formal definition for this important variable has proved difficult, due, in part, to the relatively coarse spatial resolution of the remote-sensing data used for population estimation. The advent of airborne Light Detection And Ranging (LiDAR) sensors for measuring elevation at fine spatial resolutions has provided new opportunities for considering the three-dimensional nature of living space in urban environments and for improving small-area population estimations. In this study, we assess the potential of fine-spatial-resolution LiDAR measurements (1 m) coupled with automated techniques for building extraction and land-use classification. The study seeks to provide an answer to the question: what level of information extracted from fine-spatial-resolution LiDAR and aerial photographs can be realistically translated into improved small-area population estimation? This question is addressed through a comparative study of up to seven linear models with building count, building area and/or building volume as explanatory variables at one of two land-use levels: single-family dwelling, multi-family dwelling and other types, versus residential and other types. Results show that, while building volume fits more naturally the population figures, it also represents the most challenging variable to measure by automated means. Because of this, a simple model expressed in terms of residential-building counts results in more reliable population estimates.

Fully Constrained Linear Spectral Unmixing: Analytic Solution Using Fuzzy Sets

The linear mixture model is a convenient way to describe image pixels as a linear combination of pure spectra - termed end-members. The fractional contribution from each end-member is calculated through inversion of the linear model. Despite the simplicity of the model, a nonnegativity constraint that is imposed on the fractions leads to an unmixing problem for which it is hard to find a closed analytical solution. Current solutions to this problem involve iterative algorithms, which are computationally intensive and not appropriate for unmixing large number of pixels. This paper presents an algorithm to build fuzzy membership functions that are equivalent to the least square solution of the fully constrained linear spectral unmixing problem. The efficiency and effectiveness of the proposed solution is demonstrated using both simulated and real data.

Advanced modeling of visual information processing: A multi-resolution directional-oriented image transform based on Gaussian derivatives

Abstract In this work, a multi-channel model for image representation is derived based on the scale-space theory. This model is inspired in biological insights and includes some important properties of human vision such as the Gaussian derivative model for early vision proposed by Young [The Gaussian derivative theory of spatial vision: analysis of cortical cell receptive field line-weighting profiles, General Motors Res. Labs. Rep. 4920, 1986]. The image transform that we propose in this work uses analysis operators similar to those of the Hermite transform at multiple scales, but the synthesis scheme of our approach integrates the responses of all channels at different scales. The advantages of this scheme are: (1) Both analysis and synthesis operators are Gaussian derivatives. This allows for simplicity during implementation. (2) The operator functions possess better space-frequency localization, and it is possible to separate adjacent scales one octave apart, according to Wilsons results on human vision channels. [H.R. Wilson, J.R. Bergen, A four mechanism model for spatial vision. Vision Res. 19 (1979) 19–32). (3) In the case of two-dimensional (2-D) signals, it is easy to analyze local orientations at different scales. A discrete approximation is also derived from an asymptotic relation between the Gaussian derivatives and the discrete binomial filters. We show in this work how the proposed transform can be applied to the problems of image coding, noise reduction and image fusion. Practical considerations are also of concern.

Tropical Dry Forests in the Global Picture: The Challenge of Remote Sensing-Based Change Detection in Tropical Dry Environments

Global environmental change has recently pushed the scientific community in the quest for more comprehensive spatial information on the continental biosphere. In terms of climate change, ecosystem monitoring has become one of the priorities to better understand the evolution of terrestrial carbon stocks, as well as to foster conservation policies for these carbon stocks. According to IPCC (2002), deforestation and land clearing activities, mostly from sub-tropical regions, contributed with one fifth of the greenhouse gas emission during the 1990s. The tropical dry forests are one of the most extended tropical forested ecosystems, and yet have received only recent attention from the scientific community. This ecosystem is also scarcely represented in the international protection schemes, which perhaps causes increased vulnerability of this ecosystem to the tropical fingerprint of global human development. Additionally, the climatic conditions are relatively attractive for human settlement and the ecosystem has historically supported dense agriculture activity. In megadiverse Mexico for example, these forests extend up to 60% of tropical forests, and an estimated 30% of this extent is considered as highly modified under anthropic pressure. The annual deforestation rate of the deciduous tropical dry forest in Mexico has been evaluated at around 1.4 2 %. The contribution of the latter to climate change is manifolds, including carbon emissions, increased albedo and regional hydrographic cycle alteration. Moreover, the very loss of biodiversity derived from the conversion of forest to grassland for pasture is considered as a triggering factor for future forest fires and conversion to more grassland. The monitoring and analysis of the forest distribution pattern, including phenological and anthropogenic modifications, contributes to the uneasy task of slowing down the tendency of forest loss. Remote sensing has proved a fundamental tool for such monitoring, owing to its contribution to the study and understanding of the global environment through time, and the calibration of models which help building environmental scenarios in the future.

Representing geographical objects with scale-induced indeterminate boundaries: A neural network-based data model

The degree of uncertainty of many geographical objects has long been known to be in intimate relation with the scale of its observation and representation. Yet, the explicit consideration of scaling operations when modeling uncertainty is rarely found. In this study, a neural network‐based data model was investigated for representing geographical objects with scale‐induced indeterminate boundaries. Two types of neural units, combined with two types of activation function, comprise the processing core of the model, where the activation function can model either hard or soft transition zones. The construction of complex fuzzy regions, as well as lines and points, is discussed and illustrated with examples. It is shown how the level of detail that is apparent in the boundary at a given scale can be controlled through the degree of smoothness of each activation function. Several issues about the practical implementation of the model are discussed and indications on how to perform complex overlay operations of fuzzy maps provided. The model was illustrated through an example of representing multi‐resolution, sub‐pixel maps that are typically derived from remote sensing techniques.

Extraction of buildings footprint from LiDAR altimetry data with the hermite transform

Building footprint geometry is a basic layer of information required by government institutions for a number of land management operations and research. LiDAR (light detection and ranging) is a laserbased altimetry measurement instrument that is flown over relatively wide land areas in order to produce digital surface models. Although high spatial resolution LiDAR measurements (of around 1 m horizontally) are suitable to detect aboveground features through elevation discrimination, the automatic extraction of buildings in many cases, such as in residential areas with complex terrain forms, has proved a difficult task. In this study, we developed a method for detecting building footprint from LiDAR altimetry data and tested its performance over four sites located in Austin, TX. Compared to another standard method, the proposed method had comparable accuracy and better efficiency.

Optic flow estimation using the Hermite transform

In this paper we present a spatiotemporal energy based method to estimate motion from an image sequence. A directional energy is defined in terms of the Radon projections of the Hermite transform. Radon transform provides a suitable representation for image orientation analysis, while Hermite transform describes image features locally in term of Gaussian derivatives. These operators have been used in computer vision for feature extraction and are relevant in visual system modeling. A directional response defined from the directional energy is used to estimate local motion as well as to compute a confidence matrix. This matrix provides a confidence measure for our estimate and is used to propagate the velocity information towards directions with high uncertainty. With this results, there can be applications ranging from motion compensation, and tracking of moving objects, to segmentation and video compression.

Image coding with a directional-oriented Hermite transform on a hexagonal lattice

This paper presents a novel image compression scheme based on the perceptual classification of image patterns in the Discrete Hermite Transform (DHT) domain over a roughly hexagonal sampling lattice. The DHT analyzes a signal through a set of binomial filters which approximate the Gaussian derivatives with the advantage that they are computed efficiently. In order to obtain the DHT referred to a rotated coordinate system the set of coefficients of a given order are mapped through a unitary transformation that is locally specified. Such a transformation is based on the generalized binomial functions so that the rotation algorithm is efficient too. This representation allows a perceptual classification, which is achieved by thesholding the approximation errors that are obtained under the hypotheses that the underlying pattern is a constant (0-D), an oriented structure (1-D) or a non-oriented estructure (2-D). The threshold is based on light adaptation and contrast masking properties of the human vision. Then, a compression is obtained by elimination of coefficients that are visually irrelevant.

Motion analysis and classification with directional Gaussian derivatives in image sequences

This work is intended to provide some ideas on the use of a Gaussian-derivative model for visual perception, called the Hermite transform, to extract motion information from an image sequence. Gaussian-derivative operators have long been used in computer vision for feature extraction and are relevant in visual system modeling. A directional energy is defined in terms of the 1-D Hermite transform coefficients of local projections. Each projection is described by the Hermite transform, resulting in a directional derivative analysis of the input at a given spatiotemporal scale. We demonstrate that the 1-D Hermite transform coefficients of local projections are readily computed as a linear mapping of the 3-D Hermite transform coefficients through some projecting functions. The directional response is used to detect spatiotemporal patterns that are 1-D or 2-D. Practical consideration and experimental results are also of concern.

SAR image classification with a directional-oriented discrete Hermite transform

This paper presents a novel classification scheme for SAR images based on the perceptual classification of image patterns in the Discrete Hermite Transform (DHT) domain over a roughly hexagonal sampling lattice. The DHT analyzes a signal through a set of binomial filters which approximate the Gaussian derivatives with the advantage that they are computed efficiently. In order to obtain the DHT referred to a rotated coordinate system the set of coefficients of a given order are mapped through a unitary transformation that is locally specified. Such a transformation is based on the generalized binomial functions so that the rotation algorithm is efficient too. This representation allows a perceptual classification, which is achieved by thesholding the approximation errors that are obtained under the hypotheses that the underlying pattern is a constant (0-D), an oriented structure (1-D) or a non-oriented structure (2-D). The threshold is based on light adaptation and contrast masking properties of the human vision.