Coral-CAT: A semi-automatic coral color analysis tool

Coral health assessments based on pigmentation or color rely primarily on the comparison of the coral color with well-established color references. In most cases, the assignment of color codes is done manually in a time-consuming and nonstandardized way. We present a semi-automatic image analysis tool to extract color information from coral images and assign color codes. The pipeline allows standardized and automatic extraction of the color of each pixel of a cropped coral image and assignment of the closest color code, which is extracted from the color chart included in the same picture. This feature reduces the bias due to different lightening of the images, including depth-related color distortion and variable natural or external lighting. Then, the percentage of pixels corresponding to each color code is assessed. We validated the tool against human visual assessments using a real case study where we assessed the color score of 68 images of coral fragments. Finally, we successfully assessed the color codes of underwater images of coral colonies. This tool provides an easy and objective method for the color assessment of coral images, allowing the monitoring of coral bleaching and color changes.