Aviation smoke detection method based on image recognition

Abstract

The invention relates to the technical field of aviation of image processing, and particularly discloses an aviation smoke detection method based on image recognition, which comprises the following steps of: acquiring a continuous image sequence of a plurality of visual angles in a passenger cabin, and constructing a four-dimensional space-time tensor; extracting a time sequence feature sequence along a time axis, performing directional convolution sampling according to the airflow direction parameters, obtaining space-time response values, and aggregating the space-time response values into a feature map set; generating a space weight according to a smoke diffusion physical rule, generating a time weight according to a gray level fluctuation frequency, and obtaining a feature expression after weighting; decomposing the feature expression into a spatial response diagram, obtaining three-dimensional coordinate parameters of a smoke source through spatial coordinate voting and weighted fusion, and calculating a confidence score according to the dispersion degree of candidate coordinates and an airflow path; the coordinate information is matched to passenger cabin space layout data, and situation output is generated; according to the invention, the smoke source can be accurately positioned in a strong airflow environment, and the reliability quantitative index can be output.

Images