DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Preliminary Amendment
The preliminary amendment submitted on 03/06/2025 is acknowledged. Claims 1-13 are cancelled. Claims 14-33 are pending.
Information Disclosure Statement
The information disclosure statement (IDS) submitted is considered by the examiner.
Claim Objections
Claims 14-33 are objected to because of the following informalities:
Claims 15-23 and 25-33 have a quotation mark to indicate the dependency of the limitation. It is suggested to remove the quotation mark from the preamble of each dependent claims as it is inherent that the claims further recite the limitation following.
Claims 14-15, 17-25, 27-33 do not have “and“ after the semicolon prior to the last limitation. It is suggested that the “and” to be place before the last limitation of the claims. Appropriate correction is required.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 14, 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ummernhofer et al. (NPL, “Adaptive Surface Reconstruction with Multiscale Convolutional Kernels”, 2021, hereinafter “Ummernhofer”).
(1) regarding claim 14:
As shown in fig. 1, Ummenhofer disclosed a mesh reconstruction (fig. 1, Reconstruction of a scene with varying scale. The surface triangle mesh adapts to the scale to capture details such as the fountain) method, comprising:
constructing a data structure according to three-dimensional scanning data and a preset mesh side length (page 6, 7.1 Data Generation, para. [0001], note that for sample generation, we randomly place 3D models and virtual scanners in a scene), and determining node information of a target tree node among a plurality of tree nodes comprised in the data structure (page 3, Overview, para, [0001], note that after building the octree, we extract grids at multiple resolutions starting with the leaf nodes and walking up the tree hierarchy. Also see page 3, Octree Generation and Feature Aggregation, para. [0001], note that ee assign to each input point a footprint size σ and use it to steer the subdivision of the octree such that the edge length l of the voxel containing the point is smaller than σ);
determining a level of a mesh to be extracted of the target tree node according to target data, the target data being obtained according to the three-dimensional scanning data and/or according to the data structure (page 3, Overview, para. [0003], note that we implement dual contouring for adaptive grids, which generates a triangle mesh for the zero-level set of the signed distance field. Also see Octree Generation and Feature Aggregation, para. [0002], note that after constructing the octree, we aggregate information from the input point cloud into the leaf nodes).
Ummenhofer disclosed most of the subject matter as described as above except for specifically teaching calculating a scalar field of the target tree node according to node information of a neighboring tree node in a same level as the target tree node and the node information of the target tree node; and obtaining a reconstructed mesh model according to the scalar field.
However, it would have been obvious for Ummenhofer to teach calculating a scalar field of the target tree node according to node information of a neighboring tree node in a same level as the target tree node and the node information of the target tree node (page 1, para. [0002], note that the problem of finding the 2D surface is turned into finding a 3D scalar field from which the surface can be extracted as a level set. Page 4, para. [0001], note that a scalar function that defines the importance of an input point with respect to the voxel that we want to aggregate the information into. We define the importance of a sample based on the compatibility of the scale between the point and the voxel and the distance of the point to the voxel center), and obtaining a reconstructed mesh model according to the scalar field (page 7, Results, para. [0005], note that we were able to reconstruct this dataset with the full point cloud with the GDMR baseline. Our method reconstructs the scene more than two times faster (117min) compared toGDMR(274min)).
At the time of filing for the invention, it would have been obvious to a person of ordinary skilled in the art to teach calculating a scalar field of the target tree node according to node information of a neighboring tree node in a same level as the target tree node and the node information of the target tree node; and obtaining a reconstructed mesh model according to the scalar field. The suggestion/motivation for doing so would have been in order to predict the signed and unsigned distance fields for large data sets with millions of input points and is faster and more accurate than classic energy minimization or recent learning approaches (abs.). Therefore, it would have been obvious for Ummenhofer to obtain the invention as specified in claim 14.
(2) regarding claim 19:
Ummenhofer further disclosed the method according to claim 14, wherein "determining a level of a mesh to be extracted of the target tree node according to target data" comprises:
calculating a density of scanning points comprised in the target tree node according to the node information of the neighboring tree node of the target tree node in the data structure (page 3, Overview, para. [0001], note that to steer the subdivision of the volume, we use the scale information associated with the input points or estimate the scale from the point density. After building the octree, we extract grids at multiple resolutions starting with the leaf nodes and walking up the tree hierarchy);
determining a resolution level of the target tree node according to a density of scanning points comprised in the target tree node and a second threshold (page 3, Overview, para. [0001], note that we then aggregate features from the point cloud in the grid with the highest resolution using a continuous convolution);
determining the level of the mesh to be extracted of the target tree node according to the resolution level of the target tree node (page 5, Distance Function Decode and Contouring, para. [0001], note that to extract the surface as a mesh, we use dual contouring and evaluate the signed and unsigned distance function at the voxel centers).
(3) regarding claim 20:
Ummenhofer further disclosed the method according to claim 19, wherein "calculating a density of scanning points comprised in the target tree node according to the node information of the neighboring tree node of the target tree node in the data structure" comprises:
calculating a first distance from the neighboring tree node to the target tree node according to the average coordinate in the node information of the neighboring tree node of the target tree node in the data structure and the average coordinate in the node information of the target tree node (page 5, Distance Function Decode and Contouring, para. [0001], note that we decode the signed distance value u, its spatial gradient ∇u, and the unsigned distance value v with an MLP. The input to the MLP are the features f, the relative position r, which gives the query position relative to the voxel center c and the voxel edge length l. Like r, the output of the MLP is normalized with respect to the voxel size. To find the distance values [u,v] for a voxel i at position x);
determining a contribution weight of the neighborhood tree node to the target tree node according to the first distance (fig. 3, note that to evaluate the implicit functions ˆu and ˆv for a voxel we use coordinates r that are relative to the voxel size and center and a small MLP decoder with 3 layers. Since the decoder is differentiable, we can add operations from the backward pass to the network, visualized here as transposed layers, to compute the gradient of the signed distance ∇u, giving us a decoder with a total of 6 layers of which 3 pairs share weights);
calculating the density of the scanning points comprised in the target tree node according to the contribution weight, the weighted sum in the node information of the target tree node, and the weighted sum in the node information of the neighborhood tree node (fig. 5, para. [0003], note that we mix both configurations, convolution with and convolution without normalization, to incorporate and retain information about the density from the aggregation stage. For the first convolution on each grid level in the encoder).
The proposed rejection of claims 14, 19-20, renders obvious the steps of the device claims 24, 29-30 because these steps occur in the operation of the proposed rejection as discussed above. Thus, the arguments similar to that presented above for claims 14, 19-20 are equally applicable to claims 24, 29-30.
Claim(s) 21 and 31 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ummernhofer in view of Vespa et al. (NPL, “Adaptive-resolution Octree-based Volumetric SLAM”, 2019, hereinafter “Vespa”).
(1) regarding claim 21:
Ummernhofer disclosed most of the subject matter as described as above except for specifically teaching determining a resolution level of a scanning point comprised in the three-dimensional scanning data; calculating the resolution level of the target tree node in the data structure; determining the level of the mesh to be extracted of the target tree node according to the resolution level of the scanning point and the resolution level of the target tree node.
However, Vespa disclosed determining a resolution level of a scanning point comprised in the three-dimensional scanning data (page 7, Qualitative Evaluation, para. [0001], note that all the recorded sequences we have simulated a realistic scanning scenario, where first the scene is observed closely and then the camera slowly moves away to scan other parts of the environment); calculating the resolution level of the target tree node in the data structure (page 5, Upward Propagation, para. [0001], note that we fuse information at the finest resolution possible and up-propagate. We then render the same frame at progressively coarser voxel resolution via ray-casting, obtaining consistent results); determining the level of the mesh to be extracted of the target tree node according to the resolution level of the scanning point and the resolution level of the target tree node (page 7, Qualitative Evaluation, para. [0001], note that Figures 1 and 7 compare the output of our novel re construction pipeline against a traditional single-resolution system. Both meshes are extracted via marching cubes [4] at the maximum available resolution).
At the time of filing for the invention, it would have been obvious to a person of ordinary skilled in the art to teach determining a resolution level of a scanning point comprised in the three-dimensional scanning data; calculating the resolution level of the target tree node in the data structure; determining the level of the mesh to be extracted of the target tree node according to the resolution level of the scanning point and the resolution level of the target tree node. The suggestion/motivation for doing so would have been in order to an efficient octree structure which supports multi-resolution rendering allowing for online frame-to-model alignment. (abs.). Therefore, it would have been obvious to combine Ummenhofer with Vespa to obtain the invention as specified in claim 21.
The proposed rejection of claim 21, renders obvious the steps of the device claim 31 because these steps occur in the operation of the proposed rejection as discussed above. Thus, the arguments similar to that presented above for claim 21 is equally applicable to claim 31.
Claim(s) 22-23 and 32-33 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ummernhofer in view of Vidal et al. (NPL, “A Progressive Approach to Scalar Field Topology”, 2021 hereinafter “Vidal”).
(1) regarding claim 22:
Ummernhofer disclosed most of the subject matter as described as above except for specifically teaching calculating a local scalar field of the target tree node according to the node information of the neighboring tree node in the same level as the target tree node and the node information of the target tree node; determining a global scalar field of the same level according to the local scalar fields of all target tree nodes in the same level; obtaining the reconstructed mesh model by extracting a zero isosurface according to the global scalar fields of different levels.
However, Vidal disclosed calculating a local scalar field of the target tree node according to the node information of the neighboring tree node in the same level as the target tree node and the node information of the target tree node (page 4, 2.1 Input Data, para. [0001], note that the scalar values are given at the vertices of M and are linearly interpolated on the other simplices (with barycentric coordinates). f is assumed to be injective on the vertices of M (i.e., each vertex has a distinct f value)); determining a global scalar field of the same level according to the local scalar fields of all target tree nodes in the same level (page 7, 3.3 Topologically Invariant Vertices, para. [0001], note that the input edge-nested triangulation hierarchy H yields a hierarchy of PL scalar fields, such that each old vertex v maintains by construction its scalar value); obtaining the reconstructed mesh model by extracting a zero isosurface according to the global scalar fields of different levels (page 14, fig. 15, note that Fig. 15. Progressive persistence diagrams (saddle-maximum pairs, from left to right) of the CTscan of a foot (leftmost: isosurface), at a few steps of the computation).
At the time of filing for the invention, it would have been obvious to a person of ordinary skilled in the art to teach calculating a local scalar field of the target tree node according to the node information of the neighboring tree node in the same level as the target tree node and the node information of the target tree node; determining a global scalar field of the same level according to the local scalar fields of all target tree nodes in the same level; obtaining the reconstructed mesh model by extracting a zero isosurface according to the global scalar fields of different levels. The suggestion/motivation for doing so would have been in order to enable the definition of efficient coarse-to-fine topological algorithms, which leverage fast update mechanisms for ordinary vertices and avoid computation for the topologically invariant ones; and control of the execution time of complete topological pipelines as well as previews of the topological features found in a dataset, with progressive updates delivered within interactive times (abs.). Therefore, it would have been obvious to combine Ummenhofer with Vidal to obtain the invention as specified in claim 22.
(2) regarding claim 23:
Ummernhofer disclosed most of the subject matter as described as above except for specifically teaching determining the neighboring tree node in the same level as the target tree node; determining a target plane passing through the neighborhood tree node according to the average coordinate and the average normal in the node information of the neighborhood tree node, and calculating a directed distance from the target tree node to the target plane; calculating the local scalar field of the target tree node according to the node information of the neighborhood tree node, the node information of the target tree node and the directed distance.
However, Vidal disclosed determining the neighboring tree node in the same level as the target tree node (page 11, 6.1 Progressive Data Representation, para. [0001], note that computation(Section5)progressively refines an estimation of the output, by efficiently updating at each new hierarchy level i.); determining a target plane passing through the neighborhood tree node according to the average coordinate and the average normal in the node information of the neighborhood tree node, and calculating a directed distance from the target tree node to the target plane (page 11, 6.1 Progressive Data Representation, para. [0003], note that for each level i, we measure the L2-Wasserstein distance. We normalize this distance by dividing it by W2. Then, along the hierarchy H, this normalized distance progresses from 1 to 0 for all datasets); calculating the local scalar field of the target tree node according to the node information of the neighborhood tree node, the node information of the target tree node and the directed distance (see fig. 7, note that fig. 7. Important properties of edge-nested triangulations, enabling fast updates of local topological information. (a) Left: From one hierarchy level (i) to the next (i þ 1), edge-nested triangulations preserve the local structure of the link of an old vertex v (red sphere). In particular, there exists a one-to-one mapping Ci between the vertices and the edges (red arrows). (c) Right: A new vertex v which is monotonic with v0 and v1 being respectively the lowest and highest vertex of the edge (v0,v1)).
At the time of filing for the invention, it would have been obvious to a person of ordinary skilled in the art to teach determining the neighboring tree node in the same level as the target tree node; determining a target plane passing through the neighborhood tree node according to the average coordinate and the average normal in the node information of the neighborhood tree node, and calculating a directed distance from the target tree node to the target plane; calculating the local scalar field of the target tree node according to the node information of the neighborhood tree node, the node information of the target tree node and the directed distance. The suggestion/motivation for doing so would have been in order to enable the definition of efficient coarse-to-fine topological algorithms, which leverage fast update mechanisms for ordinary vertices and avoid computation for the topologically invariant ones; and control of the execution time of complete topological pipelines as well as previews of the topological features found in a dataset, with progressive updates delivered within interactive times (abs.). Therefore, it would have been obvious to combine Ummenhofer with Vidal to obtain the invention as specified in claim 23.
The proposed rejection of claims 22-23, renders obvious the steps of the device claims 32-33 because these steps occur in the operation of the proposed rejection as discussed above. Thus, the arguments similar to that presented above for claims 22-23 are equally applicable to claims 32-33.
Allowable Subject Matter
Claim 15-18, 25-28 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims. The following is a statement of reasons for the indication of allowable subject matter: the prior arts made of record do not teach “wherein the three-dimensional scanning data comprises scanning information of a plurality of scanning points, and constructing a data structure according to three-dimensional scanning data and a preset mesh side length, and determining node information of a target tree node among a plurality of tree nodes comprised in the data structure comprises: determining a side length of a cubic space occupied by a scanned object; based on the three-dimensional scanning data, constructing the data structure with an octree shape according to the side length of the cubic space occupied by the scanned object and the preset mesh side length, wherein the data structure comprises a plurality of levels, each of the plurality of levels comprises a plurality of tree nodes with a cubic shape; determining a tree node comprising at least one scanning point from all tree nodes comprised in a preset level of the data structure as the target tree node; calculating the node information of the target tree node according to scanning information of the scanning point comprised in the target tree node”, as recited in claims 15 & 25 and “wherein determining a level of a mesh to be extracted of the target tree node according to target data comprises: calculating a curvature of each scanning point in the three-dimensional scanning data; determining a resolution level of each scanning point according to the curvature of each”, as recited in claims 18 & 28. Claims 16-17 and 26-27 depend on the claims 15 and 25 respectively.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Lee et al. (EP 1 574 996 A2) disclosed a method and apparatus for encoding and/or decoding depth image-based representation (DIBR) data are provided. The encoding method includes: converting 3-dimensional (3D) volume data into adjustable octree data with predetermined labels given to nodes; by referring to the labels, encoding nodes of the adjustable octree from the root node to leaf nodes by a modified breadth-first search (BFS) method allocating priorities among children nodes; and generating a bitstream with predetermined header information and encoded node data.
Any inquiry concerning this communication or earlier communication from the examiner should be directed to Hilina K Demeter whose telephone number is (571) 270-1676.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, King Y. Poon could be reached at (571) 270- 0728. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about PAIR system, see http://pari-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/HILINA K DEMETER/Primary Examiner, Art Unit 2617
Prosecution Insights