Image processing method, image processing apparatus and computer readable storage medium

An image processing device, comprising:

• a memory configured to store computer readable instructions;
• and
• a processor configured to run the computer readable instructions to enable the image processing device perform:
• acquiring a first image and a second image captured for a same object;
• and
• determining depth information of the object in the first image or the second image according to the first image and the second image,
• wherein the determining the depth information of the object in the first image or the second image according to the first image and the second image comprises:
• obtaining the depth information by inputting the first image and the second image into a trained neutral network,
• the neutral network includes a convolution network and a deconvolution network,
• the obtaining the depth information by inputting the first image and the second image into the trained neutral network comprises:
• inputting the first image and the second image into the convolution network,
• taking an output of the convolution network as an input of the deconvolution network,
• outputting a depth map by processing of the deconvolution network, and determining the depth information based on the depth map,
• the inputting the first image and the second image into the convolution network comprises:
• inputting a spliced image into the convolution network, wherein the spliced image is formed by splicing the first image and the second image on channels,
• the convolution network includes n convolution layers;
• the deconvolution network includes n deconvolution layers;
• and
• the taking the output of the convolution network as the input of the deconvolution network comprises:
• a feature map outputted by an ith convolution layer in the convolution network and a feature map outputted by an (n-i)th deconvolution layer in the deconvolution network are spliced on channels to form a spliced feature map, the spliced feature map is taken as an input of an (n-i+1)th deconvolution layer among the n deconvolution layers;
• and i is an integer greater than or equal to 1 and less than or equal to n−1.

A non-transitory computer readable storage medium, used for storing computer readable instructions, when the computer readable instructions are executed by the computer, the computer performs:

• acquiring a first image and a second image captured for a same object;
• and
• determining depth information of the object in the first image or the second image according to the first image and the second image,
• wherein the determining the depth information of the object in the first image or the second image according to the first image and the second image comprises:
• obtaining the depth information by inputting the first image and the second image into a trained neutral network,
• the neutral network includes a convolution network and a deconvolution network,
• the obtaining the depth information by inputting the first image and the second image into the trained neutral network comprises:
• inputting the first image and the second image into the convolution network,
• taking an output of the convolution network as an input of the deconvolution network,
• outputting a depth map by processing of the deconvolution network, and determining the depth information based on the depth map,
• the inputting the first image and the second image into the convolution network comprises:
• inputting a spliced image into the convolution network, wherein the spliced image is formed by splicing the first image and the second image on channels,
• the convolution network includes n convolution layers;
• the deconvolution network includes n deconvolution layers;
• and
• the taking the output of the convolution network as the input of the deconvolution network comprises:
• a feature map outputted by an ith convolution layer in the convolution network and a feature map outputted by an (n-i)th deconvolution layer in the deconvolution network are spliced on channels to form a spliced feature map, the spliced feature map is taken as an input of an (n-i+l)th deconvolution layer among the n deconvolution layers;
• and i is an integer greater than or equal to 1 and less than or equal to n−1.