Human image acquisition method applying binocular recognition

The invention discloses a portrait acquisition method applying binocular recognition. The portrait acquisition method comprises the following steps: acquiring different illumination images PXn and PYn through two cameras; carrying out the illumination normalization processing on the different illumination images PXn and PYn, and constructing an illumination reference image PX'n and an image PY'n; carrying out preprocessing and cognitive determination on the image PX'n and the image PY'n, and confirming the targeted cognitive features of the collected object based on the deep learning cognitive features of the images; carrying out parallax calculation according to a binocular stereo imaging principle and obtaining a point cloud picture; constructing a three-dimensional coordinate space according to the targeted cognitive features of the collected object and the point cloud atlas; and judging the recognition degree precision requirement and the error requirement of the three-dimensional coordinate space. Accord