Julio Martín-Herrero

Anisotropic Diffusion in the Hypercube

Denoising of hyperspectral images in the domain of imaging spectroscopy by vectorial anisotropic diffusion is addressed. Anisotropic diffusion has shown to be a powerful denoising technique with many applications in several fields of image processing, and in the recent years, some significant advances have been published. However, these had not yet been specifically adapted to hyperspectral imagery. This paper reviews recent advances in anisotropic diffusion for multivalued images, analyzes their application to hyperspectral images, and proposes a new diffusion method which takes advantage of the recent improvements and conforms to the specificities of hyperspectral remote sensing. Some examples are provided using both a noisy image and a clean image with added noise

Effectiveness of a varied assemblage of seed dispersers of a fleshy‐fruited plant

Disperser effectiveness is the number of new plants resulting from the activity of one disperser relative to other dispersers or to nondispersed seeds. Effectiveness remains largely uninvestigated due to the complexity of its measurement. We measured the effectiveness of seed dispersers (Larus michahellis, Turdus merula, and Oryctolagus cuniculus) of the shrub Corema album (Empetraceae) using a simulation model of the recruitment process that was parameterized with field data of seed dispersal, predation, and seedling emergence and validated with independent data on seedling density. The model allows tracking the fate of seeds dispersed by each animal and estimating, for the first time, disperser effectiveness as seedlings per square meter contributed by each disperser. It also allows quantifying the relative importance of different recruitment processes in determining the quantity and spatial distribution of recruitment. Larus michahellis was the most effective disperser in two of the three habitats studied, contributing 3-125 times more than the other two species, whose lower effectiveness depended mostly on deposition patterns (T. merula) or deleterious effects on seedling emergence (O. cuniculus). The dependence of the plant on each disperser differed between habitats and was the greatest in sparse scrub, where recruitment depended almost exclusively on gulls (90%). Quantity and quality of dispersal were not correlated; quality was a better predictor of disperser effectiveness. Seedling emergence was the most crucial process in determining both the spatial pattern of recruitment among microhabitats (99.8% of variance explained) and the quantity of recruitment within microhabitats (43-83%). A sensitivity analysis showed that increasing seed dispersal improved the recruitment for all dispersers when there was no competition for fruits. However, with limited fruits, increased dispersal of lower quality dispersers reduced overall recruitment. Our results show important differences in effectiveness among dispersers and illustrate the different influences of the components of effectiveness, which varied depending not only on the disperser but also on the circumstances (e.g., type of habitat).

Hyperspectral image analysis for precision viticulture

We analyze the capabilities of CASI data for the discrimination of vine varieties in hyperspectral images. To analyze the discrimination capabilities of the CASI data, principal components analysis and linear discriminant analysis methods are used. We assess the performance of various classification techniques: Multi-layer perceptrons, radial basis function neural networks, and support vector machines. We also discuss the trade-off between spatial and spectral resolutions in the framework of precision viticulture.

Hybrid object labelling in digital images

The application of a technique for labelling connected components based on the classical recursive technique is studied. The recursive approach permits labelling, counting, and characterizing objects with a single pass. Its main drawback lies on its very nature: Big objects require a high number of recursive calls, which require a large stack to store local variables and register values. Thus, the risk of stack overflow imposes an impractical limit on image size. The hybrid alternative combines recursion with iterative scanning and can be directly substituted into any program already using the recursive technique. I show how this alternative drastically reduces the number of consecutive recursive calls, and thus the required stack size, while improving overall performance. The method is tested on sets of uniform random binary images and binary images with a random distribution of overlapping square blocks. These test sets provide insight on the adequacy of the algorithm for different applications. The performance of the proposed technique is compared with the classical recursive technique and with an iterative two-pass algorithm using the Union-Find data structure, and the results show an overall increase of speed. The performance of the algorithm in real world machine vision applications is also shown.

GPU Geocorrection for Airborne Pushbroom Imagers

We present a method for real-time geocorrection of images from airborne pushbroom sensors using the hardware acceleration and parallel computing characteristics of modern graphics processing units. This allows very fast geocorrection without accuracy loss with respect to traditional direct methods, with very little computational load for the central processor. Our method is based on a projective texture technique originally developed for fast shadow rendering. By combining the image data with inertial navigation and positioning ancillary data, we correct the image to a digital terrain model (DTM) and produce a geocorrected and georeferenced image. The method works independently of the number of channels of the sensor. Results with an ultralight hyperspectral system show that the speed achieved with standard hardware for a 1-m grid DTM gives better than real-time performance. This allows in-flight inspection of the geocorrected image during acquisition, for more efficient coverage of large target areas. In regard to accuracy, differences with standard ray tracing direct geocorrection remained subpixel.

Anisotropic Inpainting of the Hypercube

Airborne pushbroom sensors produce images by acquiring scenes on a line-per-line basis. Depending on the motions of the aircraft carrying the sensor, the line integration time, and the targeted spatial resolution, missing areas may appear on geocorrected images. Missing pixels in geocorrected images are usually tackled by means of interpolation methods, such as nearest neighbor, but these cause visible artifacts that affect the visual quality of the result and also the performance of processing methods working on geocorrected images. We propose the use of an anisotropic diffusion inpainting method specifically devised for hyperspectral images, show some extreme examples, and discuss its convenience.

Alien plant monitoring with ultralight airborne imaging spectroscopy.

Effective management of invasive plants requires a precise determination of their distribution. Remote sensing techniques constitute a promising alternative to field surveys and hyperspectral sensors (also known as imaging spectrometers, with a large number of spectral bands and high spectral resolution) are especially suitable when very similar categories are to be distinguished (e.g. plant species). A main priority in the development of this technology is to lower its cost and simplify its use, so that its demonstrated aptitude for many environmental applications can be truly realized. With this aim, we have developed a system for hyperspectral imaging (200 spectral bands in the 380–1000 nm range and circa 3 nm spectral resolution) operated on board ultralight aircraft (namely a gyrocopter), which allows a drastic reduction of the running costs and operational complexity of image acquisition, and also increases the spatial resolution of the images (circa 5–8 pixels/m2 at circa 65 km/h and 300 m height). The detection system proved useful for the species tested (Acacia melanoxylon, Oxalis pes-caprae, and Carpobrotus aff. edulis and acinaciformis), with user’s and producer’s accuracy always exceeding 90%. The detection accuracy reported corresponds to patches down to 0.125 m2 (50% of pixels 0.5×0.5 m in size), a very small size for many plant species, making it very effective for initial stages of invasive plant spread. In addition, its low operating costs, similar to those of a 4WD ground vehicle, facilitate frequent image acquisition. Acquired images constitute a permanent record of the status of the study area, with great amount of information that can be analyzed in the future for other purposes, thus greatly facilitating the monitoring of natural areas at detailed spatial and temporal scales for improved management.

Efficiency of Semi-Implicit Schemes for Anisotropic Diffusion in the Hypercube

Semi-implicit schemes have been recently shown to speed up nonlinear diffusion in hyperspectral images while increasing the accuracy of subsequent classifiers in thematic mapping. Here, we show how semi-implicit schemes can be used to implement a truly anisotropic diffusion method for hyperspectral images, and we test the performance of different implementations in terms of computational overhead, speed, numerical accuracy, and thematic mapping performance. In addition, truly anisotropic trace-based diffusion formulations, besides a more precise steering of the diffusion processes, also allow implementation by means of local oriented Gaussian masks. We show how the implementations with the highest numerical accuracy can be also the simplest and fastest while still increasing the classification performance.

Accurate Implementation of Anisotropic Diffusion in the Hypercube

Anisotropic diffusion in the hypercube is a powerful image restoration and enhancement tool for hyperspectral imagery. Processing hyperspectral images requires special attention to accuracy, in order to preserve the high spectral precision that characterizes this kind of data. Here, we propose alternative implementations of the method with specific regard to the numerical accuracy of the solution in strict sense. We propose and describe how to apply rotationally invariant finite differences and a local filtering scheme. We also test and discuss their performance on several hyperspectral images.

Comments on “A New Operational Method for Estimating Noise in Hyperspectral Images”

In their paper, Gao estimate the additive noise in hyperspectral images by computing the standard deviation of regression residuals. They perform multiple linear regression within spectrally homogeneous regions, which they obtain by means of an ldquoobject-seekingrdquo algorithm. This letter shows that the segmentation algorithm is not correct, discusses how this affects the noise estimate, and proposes a correct algorithm that, additionally, decreases the computational load of the method.