WACODI: A generic algorithm to derive the intrinsic color of natural waters from digital images

Abstract

This document presents the WAter COlor from Digital Images (WACODI) algorithm, which extracts the color of natural waters from images collected by low-cost digital cameras, in the context of participatory science and water quality monitoring. SRGB images are converted to the CIE XYZ color space, undergoing a gamma expansion and illumination correction that includes the specular reflection at the air-water interface.The XYZ values obtained for each pixel of the image are converted to chromaticity coordinates and Hue color angle (a w ), which is a measure of color. Based on the distributions of a w in sub-sections of the image, an approximation of the intrinsic color of the water is obtained. This algorithm was applied to images acquired in 2013 during two field campaigns in Northern Europe. The Hue color angles were derived from hyperspectral measurements above and below the surface, carried out simultaneously with image acquisition. When for each station a specific illumination correction was applied, based on the corresponding hyperspectral data, a good fit (r 2 5 0.93) was obtained between the image and the spectra Hue color angles (slope 5 0.98, intercept 5 20.03). When a more generic illumination correction was applied to the same images, based on the sky conditions at the time of the image acquisition (either overcast or sunny), a slightly inferior, but still satisfactory fit resulted. Results on the application of the WACODI algorithm to the first images collected by the public via the smartphone application or "APP," developed within the European FP7 Citclops, are presented at the end of this study.

Citizen Bio-Optical Observations from Coast- and Ocean and Their Compatibility with Ocean Colour Satellite Measurements

“…Pre-requisites for a relation of citizen data to space-borne remote sensing were accomplished by thorough testing of the relationship of citizen tools to sensitive professional equipment and quality control. This resulted in a very good compliance for water colour (R 2 = 0.98), transparency (R 2 = 0.96), and fluorescence measurements (R 2 = 0.97) [26,28,30]. Quality procedures as applied for Citclops/EyeOnWater tools are implemented in an automated way wherever possible.…”

“…This will increase the awareness of the strong and weak points of these "new" observations and their actual use by scientists and water managers. In the EU Citclops project, this quality control and standardization has been taken as a corner stone, which has resulted in numerous articles with many more in the pipeline [26][27][28][29][30]37], and demonstrates the high quality of data derived with these tools as compared to standard in situ or laboratory measurements (water colour R 2 = 0.98; transparency R 2 = 0.96; Chlorophyll a fluorescence R 2 = 0.97) [26,28,30].…”

“…Because colour can be quantified accurately by citizen tools, both by visual comparison with the Forel-Ule scale or directly from digital images [26], research was initiated in the Citclops project to develop a new remote sensing colour product from satellites. Van der Woerd and Wernand [8] instrument.…”

“…Forel-Ule scale or directly from digital images [26], research was initiated in the Citclops project to develop a new remote sensing colour product from satellites. Van der Woerd and Wernand [8] demonstrated that colour, expressed mainly by the hue angle (α), can be derived accurately and consistently from Sea-viewing Wide Field-of-view Sensor (SeaWiFS), Moderate Resolution Imaging Spectroradiometer (MODIS), MEdium Resolution Imaging Spectrometer (MERIS), and Ocean and Land Colour Instrument (OLCI) data.…”

“…(1) The smartphone app EyeOnWater-Colour to measure water colour [25,26], based on the well described Forel-Ule colour comparator scale [27], (2) The KdUINO, which is an underwater buoy-based light chain to measure the attenuation of light throughout the water column [28], (3) The TrandiCam (TRANsparency unDerwater Index based on Citizen cAMera pictures) to assess underwater visibility as a measure of water transparency through underwater photographs [29], (4) And the SmartFluo as a novel smartphone based fluorometer to measure algal biomass by proxy…”