Quarternion color image processing as an alternative to classical grayscale conversion approaches for pest detection using yellow sticky traps

Abstract

Efficient detection of pests in different types of crops continues to be on today’s standards a difficult task. In order to address this problem, the implementation of an Integrated Pest Management (IPM) system involving the detection and classification of insects (pests) is essential for intensive production systems. Traditionally, this has been done by placing hunting traps and later manually counting and identifying the insects found. This has proven to be a very time-consuming and expensive process. Here’s where it enters image processing, a method that in the last few years has demonstrated to be a feasible solution to the problem. However, most of the related works with good results mostly rely on images taken from traps placed in greenhouses making the processing a bit easier given the low insect saturation of the traps, which is related to how controlled is the environment in such places. When working with the same task in fields the degree of difficulty increases exponentially given the influence of opposite conditions to the ones mentioned before. This work describes a new approach to the task, by using color image processing with quaternions. The methods proposed here provide a way to extract edge maps from images of yellow sticky traps without losing the spectral relation of the channels composing the image. As a result, all insects in the image are correctly delineated regardless of their color, size and intensity. This allows for more accurate pest detection because it is possible to discriminate and identify different types of insects. The application of this approach was compared with other methods proposed in several papers, showing promising results with much thicker and better closed edges.