阅读说明：本技术 形状推断装置、形状推断方法、程序以及记录介质 (Shape estimation device, shape estimation method, program, and recording medium ) 是由 顾磊 沈思杰 于 2019-12-05 设计创作，主要内容包括：一种在推断地面三维形状时可以提高实时性的形状推断装置。处理部生成具有多个格点的栅格；获取摄像装置所拍摄的第一摄像图像及第二摄像图像；假定多个格点中第一格点的多个高度；就假定的多个高度,根据摄像装置的三维位置及摄像装置的摄像方向,导出第一格点投影于第一摄像图像的第一假定投影位置和第一格点投影于第二摄像图像的第二假定投影位置；就假定的多个高度,导出第一假定投影位置和第二假定投影位置的相似度；根据相似度,推断第一格点的高度；根据第一格点的高度,推断摄像范围的三维形状。(A shape estimation device capable of improving real-time performance when estimating a three-dimensional shape of a ground surface. The processing unit generates a grid having a plurality of grid points; acquiring a first camera image and a second camera image shot by a camera device; assuming a plurality of heights of a first grid point in the plurality of grid points; deriving a first assumed projection position where the first grid point is projected on the first captured image and a second assumed projection position where the first grid point is projected on the second captured image, based on the three-dimensional position of the imaging device and the imaging direction of the imaging device, for the assumed plurality of heights; deriving a similarity of the first and second assumed projection positions for the assumed plurality of heights; deducing the height of the first lattice point according to the similarity; a three-dimensional shape of the imaging range is estimated from the height of the first grid point.)

A shape estimation device for estimating a three-dimensional shape from a plurality of captured images of a flying object,

includes a processing unit for performing the three-dimensional shape estimation-related processing,

the processing unit generates a grid having a plurality of grid points in an imaging range in which an imaging device provided in the flying object performs imaging;

acquiring a first camera image and a second camera image shot by the camera device;

acquiring the three-dimensional position of the camera device and the camera direction information of the camera device when the first camera image and the second camera image are shot;

assuming a plurality of heights for a first grid point of the plurality of grid points;

deriving a first assumed projection position where the first grid point is projected on the first captured image and a second assumed projection position where the first grid point is projected on the second captured image, from the three-dimensional position of the imaging device and the imaging direction of the imaging device, for the assumed plurality of heights;

deriving a similarity of the first assumed projection position in the first captured image and the second assumed projection position in the second captured image for a plurality of assumed heights;

inferring a height of the first grid point from the similarity of the assumed plurality of heights; and

and deducing the three-dimensional shape of the imaging range according to the deduced height of the first grid point.

The shape inference apparatus of claim 1,

the processing part generates sparse point cloud data according to the first camera image and the second camera image;

generating surface data according to the sparse point cloud data;

deriving an initial height of the first grid point according to the face data;

the height of the first grid point is assumed in order from the initial height.

The shape inference apparatus of claim 2,

the first camera image is one or more existing camera images,

the second camera image is more than one latest camera image,

the processing unit acquires first point cloud data generated from the first captured image;

generating second point cloud data according to the second camera image;

and generating the surface data according to the first point cloud data and the second point cloud data.

The shape inference apparatus according to any one of claims 1 through 3,

the processing unit calculates an error score indicating an error between the first and second assumed projection positions based on the similarity;

inferring a height of the first grid point from the error score.

The shape inference apparatus of claim 4,

the processing unit calculates the error score based on a distance between the first grid point and an adjacent second grid point in the grid.

The shape inference apparatus of claim 4 or 5,

the processing portion calculates the error score according to smoothness between the first lattice point and a second lattice point in the grid.

The shape inference apparatus of any one of claims 1 through 6,

the first lattice point is all lattice points in the grid.

The shape inference apparatus of any one of claims 1 through 6,

the first lattice point is a partial lattice point in the grid.

The shape inference apparatus according to any one of claims 1 through 8,

the processing unit sets the interval between each point of the grid.

A shape estimation method for estimating a three-dimensional shape from a plurality of captured images of a flying object, the method comprising:

generating a grid having a plurality of grid points in an imaging range in which an imaging device provided in the flying object performs imaging;

acquiring a first camera image and a second camera image shot by the camera device;

acquiring the three-dimensional position of the camera device and the camera direction information of the camera device when the first camera image and the second camera image are shot;

assuming a plurality of heights for a first grid point of the plurality of grid points;

deriving a first assumed projection position where the first grid point is projected on the first captured image and a second assumed projection position where the first grid point is projected on the second captured image, from the three-dimensional position of the imaging device and the imaging direction of the imaging device, for the assumed plurality of heights;

deriving a similarity of the first assumed projection position in the first captured image and the second assumed projection position in the second captured image for a plurality of assumed heights;

a step of deducing the height of the first lattice point from the similarity of the assumed plurality of heights; and

and deducing the three-dimensional shape of the imaging range according to the deduced height of the first grid point.

The shape inference method of claim 10,

the step of assuming a plurality of heights of the first grid point comprises the steps of:

generating point cloud data according to the first camera image and the second camera image;

generating surface data according to the point cloud data;

deriving an initial height of the first grid point according to the face data; and

the height of the first grid point is assumed in order from the initial height.

The shape inference method of claim 11,

the first camera image is one or more existing camera images,

the second camera image is more than one latest camera image,

the step of generating point cloud data comprises:

acquiring first point cloud data generated according to the first camera image; and

generating second point cloud data according to the second camera image,

the step of generating facet data includes the steps of: and generating the surface data according to the first point cloud data and the second point cloud data.

The shape inference method according to any one of claims 10 to 12,

the step of inferring the height of the first grid point comprises the steps of:

calculating an error score representing pixel value errors of the first and second assumed projection positions according to the similarity; and

inferring a height of the first grid point from the error score.

The shape inference method of claim 13,

the step of calculating an error score comprises the steps of: and calculating the error score according to the interval between the first lattice point and the adjacent second lattice point in the grid.

The shape inference method of claim 13 or 14,

the step of calculating an error score comprises the steps of: calculating the error score according to smoothness between the first grid point and a second grid point in the grid.

The shape inference method of any one of claims 10 through 15,

the first lattice point is all lattice points in the grid.

The shape inference method of any one of claims 10 through 15,

the first lattice point is a partial lattice point in the grid.

The shape inference method of any one of claims 10 through 17,

the step of generating the grid includes the step of setting the intervals of the points of the grid.

A program for causing a shape estimation device for estimating a three-dimensional shape from a plurality of captured images captured of a flying object to execute:

generating a grid having a plurality of grid points in an imaging range in which an imaging device provided in the flying object performs imaging;

acquiring a first camera image and a second camera image shot by the camera device;

acquiring the three-dimensional position of the camera device and the camera direction information of the camera device when the first camera image and the second camera image are shot;

assuming a plurality of heights for a first grid point of the plurality of grid points;

deriving a first assumed projection position where the first grid point is projected on the first captured image and a second assumed projection position where the first grid point is projected on the second captured image, from the three-dimensional position of the imaging device and the imaging direction of the imaging device, for the assumed plurality of heights;

deriving a similarity of the first assumed projection position in the first captured image and the second assumed projection position in the second captured image for a plurality of assumed heights;

inferring a height of the first grid point from the similarity of the assumed plurality of heights; and

and deducing the three-dimensional shape of the imaging range according to the deduced height of the first grid point.

A recording medium that is a computer-readable recording medium having a program recorded thereon, the program causing a shape estimation device that estimates a three-dimensional shape from a plurality of captured images of a flying object to execute:

generating a grid having a plurality of grid points in an imaging range in which an imaging device provided in the flying object performs imaging;

acquiring a first camera image and a second camera image shot by the camera device;

acquiring the three-dimensional position of the camera device and the camera direction information of the camera device when the first camera image and the second camera image are shot;

assuming a plurality of heights for a first grid point of the plurality of grid points;

deriving a first assumed projection position where the first grid point is projected on the first captured image and a second assumed projection position where the first grid point is projected on the second captured image, from the three-dimensional position of the imaging device and the imaging direction of the imaging device, for the assumed plurality of heights;

deriving a similarity of the first assumed projection position in the first captured image and the second assumed projection position in the second captured image for a plurality of assumed heights;

inferring a height of the first grid point from the similarity of the assumed plurality of heights; and

and deducing the three-dimensional shape of the imaging range according to the deduced height of the first grid point.