Mine augmented reality unmanned mining working face depth estimation method

The invention discloses a mine augmented reality unmanned mining working face depth estimation method, which belongs to the technical field of image processing, and comprises the following steps: constructing a depth encoder consisting of a convolutional layer, a pooling layer, a residual error layer and a down-sampling module, extracting image shallow high-resolution feature information in an encoder network shallow layer, and obtaining an image shallow high-resolution feature information; in the deep layer of the encoder network, through stacking of convolution operation, image low-resolution feature information is extracted, and a shallow-layer high-resolution feature map is fused into a deep-layer low-resolution feature map through a hierarchical feature adjustment module; constructing a depth decoder network composed of a deconvolution layer, an up-sampling module and jump connection, and estimating the depth of the image; constructing a camera pose prediction network, and reprojecting luminosity loss to