Influence of noise and sampling rate on the discrete image representation error

Introduction: Digital registration of images is accompanied not only by an error caused by finite spatial resolution of the photo matrix, but also by the effect of noise whose contribution to the total error decreases with an increase in the aperture of the photosensors in the matrix. Thus, changing the sampling rate has the opposite effect on the sampling error and on the error caused by the noise. Purpose: Finding the optimal image sampling rate which would provide the minimum sampling error in the presence of noise. Results: We have studied how an image discrete representation error depends on the sampling frequency and noise level. The image sampling process in the presence of noise was simulated in the MATLAB environment. The dependencies of the root-mean-square deviation of the sampling error caused by spectrum truncation (decrease in the passband of the low-pass filter) and the noise component of the error on the sampling frequency were plotted. A theorem is formulated on the upper bound of the sampling theorem: when sampling a function of finite duration in the presence of noise, there is a finite minimum value of the sampling error which is determined by the shape of the spectrum of the function and the noise level. Practical relevance: It is advisable to use the research results when choosing a photomatrix by the number of pixels for recording images in the presence of noise, as well as when choosing a low-pass filter passband for primary processing of a digital image.

Evaluation of Erythema Severity in Dermatoscopic Images of Canine Skin: Erythema Index Assessment and Image Sampling Reliability

The regular monitoring of erythema, one of the most important skin lesions in atopic (allergic) dogs, is essential for successful anti-allergic therapy. The smartphone-based dermatoscopy enables a convenient way to acquire quality images of erythematous skin. However, the image sampling to evaluate erythema severity is still done manually, introducing result variability. In this study, we investigated the correlation between the most popular erythema indices (EIs) and dermatologists’ erythema perception, and we measured intra- and inter-rater variability of the currently-used manual image-sampling methods (ISMs). We showed that the EIBRG, based on all three RGB (red, green, and blue) channels, performed the best with an average Spearman coefficient of 0.75 and a typical absolute disagreement of less than 14% with the erythema assessed by clinicians. On the other hand, two image-sampling methods, based on either selecting specific pixels or small skin areas, performed similarly well. They achieved high intra- and inter-rater reliability with the intraclass correlation coefficient (ICC) and Krippendorff’s alpha well above 0.90. These results indicated that smartphone-based dermatoscopy could be a convenient and precise way to evaluate skin erythema severity. However, better outlined, or even automated ISMs, are likely to improve the intra- and inter-rater reliability in severe erythematous cases.

Comparative Analysis of the Global Forest/Non-Forest Maps Derived from SAR and Optical Sensors. Case Studies from Brazilian Amazon and Cerrado Biomes

Global-scale forest/non-forest (FNF) maps are of crucial importance for applications like biomass estimation and deforestation monitoring. Global FNF maps based on optical remote sensing data have been produced by the wall-to-wall satellite image analyses or sampling strategies. The German Aerospace Center (DLR) and the Japan Aerospace Exploration Agency (JAXA) also made available their global FNF maps based on synthetic aperture radar (SAR) data. This paper attempted to answer the following scientific question: how comparable are the FNF products derived from optical and SAR data? As test sites we selected the Amazon (tropical rainforest) and Cerrado (tropical savanna) biomes, the two largest Brazilian biomes. Forest estimations from 2015 derived from TanDEM-X (X band; HH polarization) and ALOS-2 (L band; HV polarization) SAR data, as well as forest cover information derived from Landsat 8 optical data were compared with each other at the municipality and image sampling levels. The optical-based forest estimations considered in this study were derived from the MapBiomas project, a Brazilian multi-institutional project to map land use and land cover (LULC) classes of an entire country based on historical time series of Landsat data. In addition to the existing forest maps, a set of 1619 Landsat 8 RGB color composites was used to generate new independent comparison data composed of circular areas with 5-km diameter, which were visually interpreted after image segmentation. The Spearman rank correlation estimated the correlation among the data sets and the paired Mann–Whitney–Wilcoxon tested the hypothesis that the data sets are statistically equal. Results showed that forest maps derived from SAR and optical satellites are statistically different regardless of biome or scale of study (municipality or image sampling), except for the Cerrado´s forest estimations derived from TanDEM-X and ALOS-2. Nevertheless, the percentage of pixels classified as forest or non-forest by both SAR sensors were 90% and 80% for the Amazon and Cerrado biome, respectively, indicating an overall good agreement.