An overview of GabRat edge disruption and its new extensions for unbiased quantification of disruptive camouflaging patterns using randomization technique

Disruptive colorations are camouflaging patterns that use contrasting colorations to interrupt the continuity of object’s edge and disturb the observer’s visual recognition. The GabRat method has been introduced and widely used to quantify the strength of edge disruption. The original GabRat method requires a composite image where a target object is placed on a particular background. It computes the intensities of ‘frequency components’ parallel and perpendicular to the edge direction at each edge point using Gabor filters, and summarizes the ratios of these two intensities around the perimeter of the shape. However, we found that the original GabRat method has an issue that produces false signals and biases to overestimating the GabRat value depending on the edge angle. Here, we introduce GabRat-R, which can diminish that angle dependency using Gabor filters with randomized base angles. Additionally, we developed GabRat-RR, which iteratively places a target object on a background with random positions and rotation angles to average the effects of the heterogeneity and anisotropy of background. Compared with the original GabRat, our GabRat-R and GabRat-RR programs run more efficiently using multithreading techniques. Those programs are provided as built-in features of the Natsumushi 2.0 software and available from the GitHub repository, https://github.com/mtlucanid/GabRat-R.