Airtightness Detection with Improving Inter Five Frame Difference Machine Vision

Air tightness is the most important technical indicator of vacuum equipment in industrial equipment leak detection. The commonly used traditional soap bubble method relies on manual visual inspection, and its detection accuracy and efficiency depend on human focus and operational skills. Machine vision technology is applied in leak detection and leakage point localization based on the five-frame differencing method in this thesis. Due to the partial overlap and missing edge information using traditional frame difference methods, as well as poor performance in processing high frame rate videos, an improved inter five frame difference method using Otsu and Canny edge detection is proposed. Firstly, the continuous image is differentiated and converted into a binary image. Then, Canny edge detection is performed on the intermediate frames, and the binary image and edge detection results are subjected to logical operations, median filtering, and morphological processing. Finally, the complete soap bubble target is obtained. Through experiments, high frame rate videos can be processed, and more complete soap bubble contour information can be obtained in the new algorithm, which can be used for machine vision vacuum equipment airtightness detection.