Step‐by‐step image enhancement method for PTZ‐camera based crack detection in expressways

Abstract

AbstractCracks on highway pavement surfaces are among the most critical problems affecting expressway maintenance. A pan/tilt/zoom (PTZ) camera‐based crack detection method was proposed, owing to the intelligent traffic system (ITS) development of an expressway network in China. However, the quality of the crack image decreased at distant PTZ camera distances, resulting in a low mean average precision (MVP) for crack detection, which is difficult to implement in civil engineering. A weighted superposition‐based crack enhancement algorithm for crack detection was proposed to enhance crack images collected by PTZ cameras. Compared with the poor generalization ability of using a single image enhancement method, this method combines the advantages and characteristics of multiple image enhancement methods, thereby ensuring the quality of crack images. This enhances the visual characteristics of the cracks to adapt to various illuminations, thereby improving the crack detection rate and providing scientific and technological reserves for its application in highway maintenance engineering. Further comparison tests on crack detection for pavement crack images collected from the China G4/Jingshi Highway demonstrated that the proposed method was efficient in detecting is efficiently to detect cracks, achieving 96% MAP and 94% F1 values after enhancing the crack images, thereby providing an objective and promising means of crack detection using PTZ cameras.