Stitcher

This module contains various image stitchers especially designed for the BeesBook Project.

class Stitcher(config=None, rectify=True)

Bases: object

Class to create a ‘panorama’ from two images.

Warning

This class is more like an abstract class. This Stitcher can not be used to estimate the required parameters for stitching. But if you already estimated the parameters with an other Stitcher you could use this on for stitching.

See also

• FeatureBasedStitcher
• RectangleStitcher

load_parameters(homo_left=None, homo_right=None, size_left=None, size_right=None)

Load needed parameters for stitching points, angles and images.

This function becomes handy if you calculate the parameters in an earlier stitching process and did not want to calculate the parameters again and just want to map points, angles or images which were made under the same camera setup as the earlier stitching process.

Parameters:

• homo_left (ndarray) – homography (3,3) for data from the left side to form a panorama.
• homo_right (ndarray) – homography (3,3) for data from the right side to form a panorama.
• size_left (tuple) – Size of the left image, which was used to calculate homography.
• size_right (tuple) – Size of the right image, which was used to calculate homography.

get_parameters()

Return the estimated or loaded parameters of the stitcher needed for later stitching.

With this function you could save the stitching parameters and load them later for further stitching of points and angles (see set_parameters).

Use this function if you estimated the transform and did not want to estimate the parameters again.

estimate_transform(image_left, image_right, angle_left=0, angle_right=0)

Estimate transformation/homography of the left and right images/data to form a panorama.

Return the transformation matrix for the left and right image.

Parameters:

• image_left (ndarray) – Input left image.
• image_right (ndarray) – Input right image.
• angle_left (int) – Angle in degree to rotate left image.
• angle_right (int) – Angle in degree to rotate right image.

Warning

This must be overridden by a sublcass to customize stitching.

compose_panorama(image_left, image_right)

Try to compose the given images into the final panorama.

This happens under the assumption that the image transformations were estimated or loaded before.

Parameters:

• image_left (ndarray) – Input left image.
• image_right (ndarray) – Input right image.

Returns:

panorama (stitched image)

Return type:

ndarray

map_left_points(points)

Map points from the left image to the panorama.

This happens under the assumption that the image transformations were estimated or loaded before.

Parameters:points (ndarray(float)) – List of points from left image (N,2). Returns:points mapped to panorama (N,2) Return type:ndarray

map_left_points_angles(points, angles)

Map points and angles from the left image to the panorama.

This happens under the assumption that the image transformations were estimated or loaded before.

Parameters:

• points (ndarray(float)) – List of points from left image (N,2).
• angles (ndarray) – Angles in rad (length (N,)).

Returns:

• points_mapped (ndarray) – points mapped to panorama (N,2)
• angles_mapped (ndarray) – angles mapped to panorama (N,)

map_right_points(points)

Map points from the right image to the panorama.

This happens under the assumption that the image transformations were estimated or loaded before.

Parameters:points (ndarray(float)) – List of points from right image (N,2). Returns:points mapped to panorama (N,2) Return type:ndarray

map_right_points_angles(points, angles)

Map points and angles from the right image to the panorama.

This happens under the assumption that the image transformations were estimated or loaded before.

Parameters:

• points (ndarray(float)) – List of points from right image (N,2).
• angles (ndarray) – Angles in rad (length (N,)).

Returns:

• points_mapped (ndarray) – points mapped to panorama (N,2)
• angles_mapped (ndarray) – angles mapped to panorama (N,)

class FeatureBasedStitcher(config=None, rectify=True)

Bases: bb_stitcher.stitcher.Stitcher

Class to create a feature based stitcher.

estimate_transform(image_left, image_right, angle_left=0, angle_right=0)

Estimate transformation for stitching of images based on feature matching.

Parameters:

• image_left (ndarray) – Input left image.
• image_right (ndarray) – Input right image.
• angle_left (int) – Angle in degree to rotate left image.
• angle_right (int) – Angle in degree to rotate right image.

class RectangleStitcher(config=None, rectify=True)

Bases: bb_stitcher.stitcher.Stitcher

Class to create a rectangle stitcher.

The RectangleStitcher maps selected points to an abstracted rectangle.

estimate_transform(image_left, image_right, angle_left=0, angle_right=0)

Estimate transformation for stitching of images based on ‘rectangle’ Stitching.

Parameters:

• image_left (ndarray) – Input left image.
• image_right (ndarray) – Input right image.
• angle_left (int) – Angle in degree to rotate left image.
• angle_right (int) – Angle in degree to rotate right image.