A highlight removal method for autonomous recovery of cable-free seismographs in field environments

Abstract

The cable-free seismograph is an instrument for capturing seismic wave signals, which is popularly employed in underground resource exploration tasks in complex field environments due to its high flexibility. However, the efficiency of exploration operations is low because the arrangement and recovery of instruments are done manually. Therefore, autonomously arranging and recovering cable-free seismographs will hold great significance. Nevertheless, the instrument surface will produce highlight under intense illumination in field environments, which can have a negative impact on the accuracy of segmentation and contour extraction of the instrument. It will significantly reduce the visual positioning accuracy of the instrument and ultimately fail in recovery tasks. To address this problem, we analyze the characteristics of highlight on cable-free seismographs in field environments and develop a method to remove the highlight by weighting the color features of the non-highlight pixels in the neighborhoods of highlight pixels. We realize the detection of highlight pixels based on the classical modified specular-free model and propose an image terrain map model to select the weighted neighborhoods of highlight pixels to ensure that no background pixels are involved in the process of highlight removal. The experimental results show that our method meets the requirements for contour extraction and object segmentation in the autonomous instrument recovery tasks. Furthermore, it outperforms currently existing highlight removal methods in the task of highlight removal for cable-free seismographs in field environments.