Data from: Standardizing the protocol for hemispherical photographs: accuracy assessment of binarization algorithms

Glatthorn, Jonas1; Beckschäfer, Philip1

Published Oct 30, 2015 on Dryad. https://doi.org/10.5061/dryad.s9652

Data files

Oct 30, 2015 version files 93.04 MB

Abstract

Hemispherical photography is a well-established method to optically assess ecological parameters related to plant canopies; e.g. information about ground-level light regimes and the distribution of foliage within the crown space. Interpreting hemispherical photographs involves classifying pixels as either sky or vegetation. A wide range of automatic thresholding or binarization algorithms exists to classify these photographs. The variety in methodology hampers ability to compare results across studies. To identify an optimal threshold selection method, this study assessed the accuracy of seven binarization methods implemented in software currently available for the processing of hemispherical photographs. Therefore, binarizations obtained by the algorithms were compared to reference data generated through a manual binarization of a stratified random selection of pixels. This approach was adopted from the accuracy assessment of map classifications known from remote sensing studies. Percentage correct (PC) and kappa-statistics (K) were calculated. The accuracy of the algorithms was assessed for photographs taken with automatic exposure settings (auto-exposure) and photographs taken with settings which avoid overexposure (histogram-exposure). In addition, gap fraction values derived from hemispherical photographs were compared with estimates derived from the manually classified reference pixels. All tested algorithms were shown to be sensitive to overexposure. Three of the algorithms showed an accuracy which was high enough to be recommended for the processing of histogram-exposed hemispherical photographs: "Minimum" (PC 98.8%; K 0.952), "Edge Detection" (PC 98.1%; K 0.950), and "Minimum Histogram" (PC 98.1%; K 0.947). The Minimum algorithm overestimated gap fraction least of all (11%). The overestimation by the algorithms Edge Detection (63%) and Minimum Histogram (67%) were considerably larger. For the remaining four evaluated algorithms (IsoData, Maximum Entropy, MinError, and Otsu) an incompatibility with photographs containing overexposed pixels was detected. When applied to histogram-exposed photographs, these algorithms overestimated the gap fraction by at least 180%.

Usage notes

reference_pixels_raw

The file contains the raw data of the reference pixel classification by an operator. Column description: id: unique id refering to a single reference pixel / pic: the picture where the pixel derived from originally / x, y: x- and y- coordinates of the pixles within the pixture / gvcl: One of 16 gray value classes used for the stratified sampling approach to select the reference pixels / r, g, b: the red, green and blue gray value of the pixel (8-bit) / class: classification of the pixel by a human operator

thresholds

The file contains thresholds of the hemispherical photographs which have been used in the publications. For details about the thresholding algorithms see the methodology section of the paper.

hemispherical_photographs

The processed hemispherical photographs. The original photographs were cropped and LZW-compressed to save memory but are not modified in any other way. The filenames indicate the site number and the used exposure protocol. 'auto_exp' stands for the automatic exposure settings of the camera. 'hist_exp' stands for the histogram exposure protocol introduced by Beckschäfer et al. (2013) (DOI:10.3832/ifor0957-006). The camera was a Nikon D70s DSLR camera equipped with a Sigma Circular Fisheye 4.5 mm 1:2.8 lens with a field of view of 180°.

accuracy_estimates

The file contains accuracy estimates of binarization algorithms of hemispherical photographs. Column description: pic: the name of the picture. From row 142 statistics over all pictures are given. / 'sub' refers to a subset of pictures without sites VIII and IX. 'all' refers to all pictures without exclusion. / bin: name of the algorithms. See the methodology section of the publication for details / pc: percentage correct estimate / var_pc: variance of pc / lower_pc: lower bound of the confidence interval of pc / upper_pc: upper bound of the confidence interval of pc / kap: kappa estimate / var_kap: variance of kappa / lower_kap: lower bound of the confidence interval of kappa / upper_kap: upper bound of the confidence interval of kappa