SYSTEMS AND METHODS FOR IMAGE CAPTURE
Images subject to panoramic image stitching are analyzed for suitability in three dimensional construction. Effects of translation changes between cameras are mitigated. The dioptric relationship of subject content of the images to the camera set identifies camera translation changes, and eligible camera position inputs, permissible to avoid parallax errors. Images within permissible ranges of each other are stitched to create panoramic images of subject content that a camera cannot fully capture with a single image. Registration plane selection based on translation distances between cameras, and permissible translation changes based on detected subject content depth are disclosed.
1 - 174 . (canceled)
175 . A method for stitching a panoramic image, the method comprising:
receiving two or more candidate panoramic subimages;
for each candidate panoramic subimage of the two or more candidate panoramic subimages, classifying one or more features of the candidate panoramic subimage as a subject of interest based on a segmentation of the candidate panoramic subimage;
aligning matching features corresponding to the subject of interest across the candidate panoramic subimages by warping the candidate panoramic subimages to create a stitched panoramic image, wherein warping the candidate panoramic subimages comprises warping a respective candidate panoramic subimage to a registration plane such that the matching features from each of the candidate panoramic subimages align on the registration plane.
176 . The method of claim 175 , wherein the registration plane is a base candidate panoramic subimage selected from the two or more candidate panoramic subimages.
177 . The method of claim 176 , wherein the base candidate panoramic subimage is selected by computing, for each camera pose associated with each of the candidate panoramic subimages, a translation movement distance between the camera pose and a furthest camera pose among all camera poses associated with the candidate panoramic subimages.
178 . The method of claim 177 , further comprising selecting the candidate panoramic subimage associated with the camera pose having a minimum translation movement distance to its respective furthest camera pose as the base candidate panoramic subimage.
179 . The method of claim 177 , further comprising selecting the candidate panoramic subimage associated with the camera pose closest to a median translation movement distance among all candidate panoramic subimages' camera poses.
180 . The method of claim 179 , wherein the median translation distance is an average of the distance between the two camera poses at the lateral limits of the candidate panoramic subimages.
181 . The method of claim 179 , wherein the median translation distance is an average of the distances of the two middle candidate panoramic subimages, as measured from at least one lateral limit camera pose of the candidate panoramic subimages.
182 . The method of claim 176 , wherein the base candidate panoramic subimage is selected by:
producing a homography of a particular surface across all candidate panoramic subimages; and
selecting the candidate panoramic subimage that generates homographies with the other candidate panoramic subimages with smallest rotation matrices.
183 . The method of claim 175 , wherein the base candidate panoramic subimage is selected by:
projecting the matching features of a candidate panoramic subimage into its warped candidate panoramic subimage; and
selecting the candidate panoramic subimage producing the least projection error in the warped candidate panoramic subimages.
184 . The method of claim 175 , further comprising selecting additional matching features of the candidate panoramic subimage that are no closer to a camera that captured the candidate panoramic subimages than the one or more features corresponding to the subject of interest.
185 . The method of claim 175 , further comprising validating the stitched panoramic image based on convergence of constituent lines of a geometric feature of the stitched panoramic image to a substantially similar vanishing point.
186 . A system comprising:
one or more processors; and
memory, wherein the memory stores one or more programs capable of execution by the one or more processors, and the one or more programs comprise instructions for:
receiving two or more candidate panoramic subimages;
for each candidate panoramic subimage of the two or more candidate panoramic subimages, classifying one or more features of the candidate panoramic subimage as a subject of interest based on a segmentation of the candidate panoramic subimage;
aligning matching features corresponding to the subject of interest across the candidate panoramic subimages by warping the candidate panoramic subimages to create a stitched panoramic image, wherein warping the candidate panoramic subimages comprises warping a respective candidate panoramic subimage to a registration plane such that the matching features from each of the candidate panoramic subimages align on the registration plane.
187 . The system of claim 186 , wherein the registration plane is a base candidate panoramic subimage selected from the two or more candidate panoramic subimages.
188 . The system of claim 187 , wherein the base candidate panoramic subimage is selected by computing, for each camera pose associated with each of the candidate panoramic subimages, a translation movement distance between the camera pose and a furthest camera pose among all camera poses associated with the candidate panoramic subimages.
189 . The system of claim 188 , wherein the operations further comprise selecting the candidate panoramic subimage associated with the camera pose having a minimum translation movement distance to its respective furthest camera pose as the base candidate panoramic subimage.
190 . The system of claim 188 , wherein the operations further comprise selecting the candidate panoramic subimage associated with the camera pose closest to a median translation movement distance among all candidate panoramic subimages' camera poses.
191 . The system of claim 190 , wherein the median translation distance is an average of the distance between the two camera poses at the lateral limits of the candidate panoramic subimages.
192 . The system of claim 190 , wherein the median translation distance is an average of the distances of the two middle candidate panoramic subimages, as measured from at least one lateral limit camera pose of the candidate panoramic subimages.
193 . The system of claim 187 , wherein the base candidate panoramic subimage is selected by:
producing a homography of a particular surface across all candidate panoramic subimages; and
selecting the candidate panoramic subimage that generates homographies with the other candidate panoramic subimages with smallest rotation matrices.
194 . The system of claim 186 , wherein the base candidate panoramic subimage is selected by:
projecting the matching features of a candidate panoramic subimage into its warped candidate panoramic subimage; and
selecting the candidate panoramic subimage producing the least projection error in the warped candidate panoramic subimages.
195 . The system of claim 186 , wherein the operations further comprise selecting additional matching features of the candidate panoramic subimage that are no closer to a camera that captured the candidate panoramic subimages than the one or more features corresponding to the subject of interest.
196 . The system of claim 186 , wherein the operations further comprise validating the stitched panoramic image based on convergence of constituent lines of a geometric feature of the stitched panoramic image to a substantially similar vanishing point.