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 .
Response to Amendment
The Amendment filed May 20th 2026 has been entered. Claims 1-14 and 18-21 are pending in the application. Claims 11-14 were previously allowed. Claim 15 was previously cancelled. Claims 16 and 17 are newly cancelled. Claims 18-21 are newly added. A further search has been performed to address the material amended in the newly added claims. Claim 18 is objected to as containing allowable subject matter but being dependent on a rejected base claim. Newly found references Tulsiani (NPL: Learning Shape Abstractions by Assembling Volumetric Primitives) and Paschalidou (NPL: Unsupervised Hierarchical Part-based Decomposition) were used for the newly added claims.
Response to Arguments
Applicant’s arguments, see Pg. 7-10, filed May 20th 2026, with respect to the rejections of claims 1 and 6 under 103 have been fully considered and are persuasive. Therefore, the rejections have been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng.
Because some of the same prior art was cited in the new grounds of rejection, arguments that could still apply to the new rejection are addressed below. Arguments that were not addressed were considered moot because of the new grounds of rejection.
Applicant argues that: “Zhongyang fails to fix the deficiencies of Deprelle and Li. Like Deprelle and Li, Zhongyang, fails to disclose or suggest "for each primitive, accessing a local point set using the set of query parameters and the query shape associated with the primitive" as recited in claim 1. The Office Action does not cite Zhongyang regarding the recited feature of claim 1. Applicant submits that Zhongyang does not disclose or suggest "for each primitive, accessing a local point set using the set of query parameters and the query shape associated with the primitive" as required by claim 1.”
In acknowledging that Zhongyang was previously not cited with respect to this feature, the Examiner performed further search and found that Zhongyang teaches: “Establish a space sphere s with radius r for each point p in the LiDAR point cloud, if the neighboring point of point p is in this space sphere, extracting the feature of points in the space sphere using PointNet network” (Pg. 3, Section B: Generation of Multi-scale, par. 2). Note that Zhongyang showcases this functionality in Fig. 3 on Pg. 3. Therefore, the Examiner submits that Zhongyang teaches extracting a local point set using parameters of a sphere shape associated with a primitive.
The Applicant argues that: “Pages 6-7 of the Office Action concede, "Deprelle fails to teach: ... for each local point set, determining, using a first neural network, a descriptor vector partitioned into:". For at least this reason, Applicant submits that Deprelle does not disclose or suggest the full breadth of the feature "for each local point set, determining, using a first neural network, a descriptor vector partitioned into: a primitive-update sub-vector configured to encode changes to geometric parameters of the initialized primitive including at least one of scale, orientation, or curvature, and a local descriptor sub-vector" as required by claim 1”.
The Examiner performed further search and notes that Deprelle teaches “the concatenation of the coordinates of a point from the associated elementary structure and the shape feature predicted by the shape encoder” (Pg. 4, Adjustment module, MLP adjustment). The coordinates of a point from the associated elementary structure were found to be analogous to the claimed primitive-update sub-vector and the shape feature predicted by the shape encoder was found to be analogous to the claimed local descriptor sub-vector. In an effort to prevent text duplication, the Examiner would like to direct the Applicant to Note 1A and Note 1B below for a comprehensive explanation.
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.
Claims 1, 2, 4, 5, 6, 7, 9, 10, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Deprelle et al. (NPL: Learning elementary structures for 3D shape generation and matching) in view of Stack Overflow (NPL: What is a vector in terms of machine learning?), Li et al: (NPL: Supervised Fitting of Geometric Primitives to 3D Point Clouds), Zhongyang (NPL: Classification of LiDAR Point Cloud based on Multiscale Features and PointNet; from Applicant’s IDS), Deng (NPL: PPFNet: Global Context Aware Local Features for Robust 3D Point Matching).
Regarding claim 1:
Deprelle teaches:
A method for adaptively abstracting a point cloud, the method comprising:
for each primitive (Deprelle: We seek to automatically learn a set of primitives (called “learned elementary structures”) for shape reconstruction and matching, Pg. 2, Fig. 1), accessing a local point set (Deprelle: For each k ∈ {1,...,K}, we start from an initial surface Sk on which we sample N points to obtain an initial point cloud Sk, Pg. 4, Section 3.1: Learnable elementary structures, par. 1)
for each local point set (Deprelle: The goal of the adjustment modules pk is to reconstruct the input shape by positioning each elementary structure, Pg. 4, Adjustment module), determining, using a first neural network (Deprelle: we use as shape encoder a simplified version of the PointNet network [21] used in [10, 11], Pg. 4, Section 3.2: Architecture details), a descriptor vector (Deprelle: the concatenation of the coordinates of a point from the associated elementary structure and the shape feature predicted by the shape encoder, Pg. 4, Adjustment module, MLP adjustment) partitioned into:
a primitive-update sub-vector configured to encode changes to geometric parameters (Deprelle: we consider a point translation learning module which translates independently each of the points sk,i by a learned vector tk,i, ek,i = tk,i + sk,I; Pg. 4, Section 3.1: Learnable elementary structures, par. 1) of the initialized primitive (Deprelle: we start from an initial surface Sk; Pg. 4, Section 3.1: Learnable elementary structures, par. 1) including at least one of scale, orientation, or curvature (see Note 1A),
and a local descriptor sub-vector (Deprelle: We represent each shape by a feature vector f(Z) computed by a point set encoder f, Pg. 3, Section 3: Approach; see Note 1B and Note 1C);
updating the set of primitives (Deprelle: At test time, the elementary structures are deformed by adjustment modules to create the output 3D shape, Pg. 3, Section 3: Approach) based on the primitive-update sub-vector (Deprelle: This module thus allows the network to update independently the position of each point on the surface, Pg. 4, Section 3.1: Learnable elementary structures) and
a global descriptor determined by a second neural network that is distinct from the first neural network (Deprelle: We then apply max-pooling over all point features followed by a linear layer, producing a global shape feature used as input to the adjustment modules, Pg. 4, Section 3.2: Shape encoder).
Note 1A: The Examiner considers the value ek,i analogous to the primitive-update sub vector, for the following reasons:
ek,I represents a primitive-update: The points described by ek,i are deformed points of an elementary structure: “deformed points ek,i […] The result of either module results in a set of elementary structure points ek,i = ψk(sk,i), and we write the elementary structure Ek as the union of the independently deformed or translated points” (Pg. 4, Section 3.1: Learnable elementary structures)
ek,i may be a vector: Deprelle teaches that ek,i represents 3D coordinates of the points of the primitive. However, as this data is output of the machine learning model taught by Deprelle, the Examiner submits that one of ordinary skill in the art would find it obvious that ek,i may be a vector, because Stack Overflow teaches: “vector is a general term with many uses. In this case, think of it as a list of values or a row in a table. The data structure is a 1-dimensional array; a vector of N elements is an N-dimensional vector, one dimension for each element. For instance, the input (3.14159, 2.71828, 1.618) is a vector of 3 elements, and could be represented as a point in 3-dimensional space.” (Stack Overflow, Pg. 1)
ek,i may be a sub-vector: Given that ek,i is concatenated for input into the MLP (“each adjustment module uses a multi-layer perceptron (MLP) that takes as inputs the concatenation of the coordinates of a point from the associated elementary structure”, Deprelle, Pg. 4, Adjustment module), ek,i may be described as a sub-vector to the concatenated input vector.
ek,I encodes changes to geometric parameters: Deprelle teaches that: “we consider a point translation learning module which translates independently each of the points sk,i by a learned vector tk,i, ek,i = tk,i + sk,i.” That is, ek,i is defined by changes to geometric parameters (tk,i is a vector that modifies the original geometry sk,i).
ek,i encodes changes to geometric parameters of the initialized primitive: In the equation cited above, sk,i represents the points of the initial primitive: “we start from an initial surface Sk on which we sample N points to obtain an initial point cloud Sk. We then pass each sampled point sk,i ∈ Sk for i ∈ {1,...,N} through elementary structure learning modules ψk.” (Deprelle, Pg. 4, Section 3.1: Learnable elementary structures, par. 1)
the geometric changes to ek,i include at least one of scale, orientation, or curvature: Deprelle teaches that “we consider a point translation learning module which translates independently each of the points sk,i by a learned vector tk,i”. That is, the orientation (translation) of the primitive is modified by the learned vector.
Note 1B: The Examiner considers the shape feature vector taught to be analogous to the claimed local descriptor sub-vector for the following reasons:
f(Z) is local: Each local shape is represented by a respective feature vector: “We represent each shape by a feature vector f(Z) computed by a point set encoder f”.
f(Z) is a descriptor: One of ordinary skill in the art may reasonably consider a feature vector that “represents” each shape as a descriptor.
f(Z) is a sub-vector: Given that f(Z) is concatenated for input into the MLP (“each adjustment module uses a multi-layer perceptron (MLP) that takes as inputs […] the shape feature predicted by the shape encoder”, Deprelle, Pg. 4, Adjustment module), f(Z) may be described as a sub-vector to the concatenated input vector.
Note 1C: The specification of the present application similarly teaches that: “Local point sets are also fed into a separate neural network, for instance the PointNet architecture, which extracts local codewords 141 for all point sets” (Pg. 8, ln. 6-8).
Deprelle fails to teach:
initializing, from a group of primitives comprising at least one of patches, volumetric shapes, or sparse meshes, a set of primitives from a predefined primitive library, associated with a query shape and a set of query parameters;
for each primitive, accessing a local point set using the set of query parameters and the query shape associated with the primitive;
combining the local descriptor sub-vector with a global descriptor determined by a second neural network that is distinct from the first neural network.
Li teaches:
initializing, from a group of primitives comprising at least one of patches, volumetric shapes, or sparse meshes, a set of primitives from a predefined primitive library, (Li: Our framework supports four types of primitives: plane, sphere, cylinder, and cones , Pg. 2, par. 2), associated with a query shape and a set of query parameters (Li: Primitive types and parameters, Pg. 4, Figure 3);
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Li with Deprelle. Initializing a set of primitives associated with a query shape and a set of query parameters, as in Li, would benefit the Deprelle teachings by enabling the fitting of known geometrical shapes. Furthermore, Deprelle teaches that their elementary structures (primitives) may include the primitives included in the library taught by Li: “If the elementary structures were unit squares or a unit sphere, then this equation would describe exactly the AtlasNet [11] model.” (Pg. 3, Section 3: Approach, par. 3)
Deprelle in view of Stack Overflow, Li fails to teach:
for each primitive, accessing a local point set using the set of query parameters and the query shape associated with the primitive;
combining the local descriptor sub-vector with a global descriptor determined by a second neural network that is distinct from the first neural network.
Zhongyang teaches:
for each primitive, accessing a local point set using the set of query parameters and the query shape associated with the primitive (Zhongyang: Establish a space sphere s with radius r for each point p in the LiDAR point cloud, if the neighboring point of point p is in this space sphere, extracting the feature of points in the space sphere using PointNet network, Pg. 3, Section B: Generation of Multi-scale, par. 2; see Note 1D);
Note 1D: The Examiner interprets the query shape to be a “space sphere s” and the query parameter(s) to be the “radius r”. This is because the specification teaches a similar “ball query” procedure: “local point sets 203 are constructed by a set of around for each primitive using a ball query procedure of fixed length” (Pg. 8, ln. 1-3).
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Zhongyang with Deprelle in view of Stack Overflow, Li. Accessing a local point set using the set of query parameters and the query shape associated with the primitive, as in Zhongyang, would benefit the Deprelle in view of Stack Overflow, Li teachings by enabling the system to “achieve large-scale scene classification of LiDAR point cloud data and improve classification accuracy” (Zhongyang, Pg. 2, par. 1).
Deprelle in view of Stack Overflow, Li and Zhongyang still fails to teach:
combining the local descriptor sub-vector with a global descriptor determined by a second neural network that is distinct from the first neural network.
Deng teaches:
A method for adaptively abstracting a point cloud, the method comprising:
initializing, from a group of primitives comprising at least one of patches (Deng: we will describe, PPFNet, trained on PPFs, points and normals of local patches, Pg. 2, col. 2, par. 4), volumetric shapes, or sparse meshes, a set of primitives from a predefined primitive library, (Li: Our framework supports four types of primitives: plane, sphere, cylinder, and cones , Pg. 2, par. 2), associated with a query shape and a set of query parameters (Li: Primitive types and parameters, Pg. 4, Figure 3);
for each local point set, determining, using a first neural network, a descriptor vector (Deng: Fr;; Pg. 4, Equation 6. See also Pg. 4, Fig. 3)
combining the local descriptor sub-vector with a global descriptor (Deng: This global feature is then concatenated to every local feature, Pg. 4, Network architecture) determined by a second neural network that is distinct from the first neural network (Deng: A max pooling layer then aggregates all the local features into a global one, summarizing the distinct local information to the global context of the whole fragment, Pg. 4, Network architecture).
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Deng with Deprelle in view of Stack Overflow, Li and Zhongyang. Combining the local descriptor sub-vector with a global descriptor, as in Deng, would benefit the Deprelle in view of Stack Overflow, Li and Zhongyang teachings by ensuring robustness to global transformations: “We hypothesize that an input guidance of PPF would aid the network to be more tolerant to rigid transformations. To test this, we gradually rotate fragments around z-axis to 180◦ with a step size of 30◦ and then match the fragment to the non-rotated one. As we can observe from Tab. 6, with PPFs, the feature is more robust to rotation and the ratio in matching performance of two networks opens as rotation increases.” (Deng, Pg. 8, What does adding PPF bring?, par. 2)
Regarding claim 2:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above), wherein the global descriptor is used as an input for determining the primitive-update sub-vector for the local descriptor (Deprelle: a global shape feature used as input to the adjustment modules, Pg. 4, Section 3.2: Architecture details; see Note 2A).
Note 2A: The adjustment modules generate ek,i, the primitive-update sub-vector for the local descriptor (as discussed in Note 1A). Because the global shape feature is used as an input for the adjustment module, the Examiner interprets Deprelle to teach a global descriptor used as an input for determining the primitive-update sub-vector for the local descriptor.
Regarding claim 4:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above), wherein updating the set of primitives is performed using the primitive-update sub-vector (Deprelle: This module thus allows the network to update independently the position of each point on the surface. The result of either module results in a set of elementary structure points ek,i, (Pg. 4, Section 3.1: Learnable elementary structures, par. 3; see Note 4A).
Note 4A: In Note 1A, it was discussed that the value ek,i taught by Deprelle is analogous to the primitive-update sub-vector. On Pg. 4 cited above, Deprelle teaches that an update of the primitive results in ek,i, and therefore updating the set of primitives inherently includes using the primitive-update sub-vector.
Regarding claim 5:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above), wherein at least two types of primitives are initialized by initializing at least two distinct query shapes (see Note 5A) and wherein the at least two distinct query shapes are used to learn a combination of primitives from the point cloud (Deprelle: In these cases, meshed planes or spheres can be deformed into complex 3D structures [11, 30]. We extend this line of work by proposing a technique for learning the base shapes that are further used to approximate the shapes in the collection, Pg. 3, par. 2; see Note 5A).
Note 5A: Deprelle teaches that the prior art may use “a collection of simple hand-picked parametric primitives.” (Pg. 2) For example, “We aim to learn shared elementary structures to reconstruct a set of 3D shapes […] If the elementary structures were unit squares or a unit sphere, then [Equation 1] would describe exactly the AtlasNet [11] model.” (Pg. 3, Section 3: Approach, pars. 1-3).
Regarding claim 6:
Claim 6 is substantially similar to claim 1, and is therefore rejected for similar reasons. Claim 6 contains the following notable differences from claim 1:
Claim 6 is directed towards an apparatus instead of a method. Deprelle teaches an apparatus:
An apparatus comprising:
a processor (Deprelle: We train our model on an NVIDIA 1080Ti GPU, with a 16 core Intel I7-7820X CPU (3.6GHz), Pg. 5, Section 3.3 Losses and training, Training details); and
a memory (Deprelle: We train our model […] with a 16 core Intel I7-7820X CPU (3.6GHz), 126GB, Pg. 5, Section 3.3 Losses and training, Training details; emphasis added; see Note 6A) storing instructions operative, when executed by the processor, to cause the apparatus to:
Note 6A: By listing “126GB”, Deprelle teaches that the CPU they utilize operates with or is connected to 126 gigabytes of memory.
Regarding claim 7:
Claim 7 is substantially similar to claim 2, and is therefore rejected for similar reasons. Claim 7 contains the following notable differences from claim 2:
Claim 7 is directed towards an apparatus instead of a method. Deprelle was shown to teach an apparatus in the mapping of claim 6 above.
Regarding claim 9:
Claim 9 is substantially similar to claim 4, and is therefore rejected for similar reasons. Claim 9 contains the following notable differences from claim 4:
Claim 9 is directed towards an apparatus instead of a method. Deprelle was shown to teach an apparatus in the mapping of claim 6 above.
Regarding claim 10:
Claim 10 is substantially similar to claim 5, and is therefore rejected for similar reasons. Claim 10 contains the following notable differences from claim 5:
Claim 10 is directed towards an apparatus instead of a method. Deprelle was shown to teach an apparatus in the mapping of claim 6 above.
Regarding claim 19:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above), further comprising encoding the combined local descriptor sub-vector and global descriptor (Deng: A group of MLPs are used to further fuse the global and local features into the final global-context aware local descriptor, Pg. 4, Network architecture; see Note 19A).
Note 19A: The Examiner submits that fusing the local and global features “into the final global-context aware local descriptor” by using one or more multi-layer perceptrons (MLPs) is analogous to encoding the combined local descriptor sub-vector and global descriptor.
Claims 3 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Deprelle et al. (NPL: Learning elementary structures for 3D shape generation and matching) in view of Stack Overflow (NPL: What is a vector in terms of machine learning?), Li et al: (NPL: Supervised Fitting of Geometric Primitives to 3D Point Clouds), Zhongyang (NPL: Classification of LiDAR Point Cloud based on Multiscale Features and PointNet; from Applicant’s IDS), Deng (NPL: PPFNet: Global Context Aware Local Features for Robust 3D Point Matching) and Qi et al.: (NPL: PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space).
Regarding claim 3:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above), wherein the set of primitives is initialized by sampling the point cloud (Deprelle: For each k ∈ {1, . . . , K}, we start from an initial surface Sk on which we sample N points to obtain an initial point cloud Sk, Pg. 4, par. 2).
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng fails to teach:
wherein the set of primitives is initialized by farthest point sampling the point cloud.
Qi teaches:
wherein the set of primitives is initialized by farthest point sampling the point cloud (Qi: Given input points {x1, x2, ..., xn}, we use iterative farthest point sampling (FPS) to choose a subset of points {xi1, xi2, ..., xim}, Pg. 3, par. 7).
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Qi with Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng. Having the set of primitives be initialized by farthest point sampling the point cloud, as in Qi, would benefit the Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teachings because “compared with random sampling, it has better coverage of the entire point set given the same number of centroids.” (Qi, Pg. 3, par. 7)
Regarding claim 8:
Claim 8 is substantially similar to claim 3, and is therefore rejected for similar reasons. Claim 8 contains the following notable differences from claim 3:
Claim 8 is directed towards an apparatus instead of a method. Deprelle was shown to teach an apparatus in the mapping of claim 6 above.
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Deprelle et al. (NPL: Learning elementary structures for 3D shape generation and matching) in view of Stack Overflow (NPL: What is a vector in terms of machine learning?), Li et al: (NPL: Supervised Fitting of Geometric Primitives to 3D Point Clouds), Zhongyang (NPL: Classification of LiDAR Point Cloud based on Multiscale Features and PointNet; from Applicant’s IDS), Deng (NPL: PPFNet: Global Context Aware Local Features for Robust 3D Point Matching) and Paschalidou (NPL: Unsupervised Hierarchical Part-based Decomposition).
Regarding claim 20:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above),
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng fails to explicitly teach:
wherein the set of primitives comprises a first set of primitives, and wherein the method further comprises splitting the first set of primitives into a second set of primitives and a third set of primitives.
Paschalidou teaches:
wherein the set of primitives comprises a first set of primitives, and wherein the method further comprises splitting the first set of primitives into a second set of primitives and a third set of primitives (Paschalidou: At every depth level, each of the 2d ∣ d = 0,…,D nodes is recursively split into two nodes (its children) until reaching the maximum depth. This results in a representation with various levels of detail; see Note 20A).
Note 20A: The Figure on Pg. 4 of Paschalidou showcases this functionality.
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Paschalidou with Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng. Splitting the first set of primitives into a second set of primitives and a third set of primitives, as in Paschalidou, would benefit the Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teachings by limiting the number of primitives that can be used to approximate a point cloud, thereby maintaining simplicity and system performance.
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Deprelle et al. (NPL: Learning elementary structures for 3D shape generation and matching) in view of Stack Overflow (NPL: What is a vector in terms of machine learning?), Li et al: (NPL: Supervised Fitting of Geometric Primitives to 3D Point Clouds), Zhongyang (NPL: Classification of LiDAR Point Cloud based on Multiscale Features and PointNet; from Applicant’s IDS), Deng (NPL: PPFNet: Global Context Aware Local Features for Robust 3D Point Matching) and Tulsiani (NPL: Learning Shape Abstractions by Assembling Volumetric Primitives).
Regarding claim 21:
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teaches:
The method of claim 1 (as shown above),
Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng fails to teach:
wherein the set of primitives comprises a first set of primitives, and wherein the method further comprises merging a second set of primitives with a third set of primitives to generate the first set of primitives.
Tulsiani teaches:
wherein the set of primitives comprises a first set of primitives, and wherein the method further comprises merging a second set of primitives with a third set of primitives to generate the first set of primitives (Tulsiani: We first train the network using a fixed high value of pm across primitives and later allow the network to also learn pm while also encouraging simplicity by the external parsimony reward. As shown in Figure 5, this has the effect of first using a large number of primitives and in later stages, merging them together and using fewer primitives, Pg. 5, Implementation Details, par. 2).
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to combine the teachings of Tulsiani with Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng. Merging a second set of primitives with a third set of primitives to generate the first set of primitives, as in Tulsiani, would benefit the Deprelle in view of Stack Overflow, Li, Zhongyang, and Deng teachings by limiting the number of primitives that can be used to approximate a point cloud, thereby maintaining simplicity and system performance.
Allowable Subject Matter
Claim 18 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:
Claim 18 recites: “wherein updating the set of primitives based on the primitive-update sub-vector comprises adding the primitive-update sub-vector to at least one primitive of the set of primitives.”
In Note 1A, the Examiner submitted that the value ek,i taught by Deprelle was analogous to the primitive-update sub-vector:
“ek,i may be a sub-vector: Given that ek,i is concatenated for input into the MLP (“each adjustment module uses a multi-layer perceptron (MLP) that takes as inputs the concatenation of the coordinates of a point from the associated elementary structure”, Deprelle, Pg. 4, Adjustment module), ek,i may be described as a sub-vector to the concatenated input vector.”
“ek,I encodes changes to geometric parameters: Deprelle teaches that: “we consider a point translation learning module which translates independently each of the points sk,i by a learned vector tk,i, ek,i = tk,i + sk,i.” That is, ek,i is defined by changes to geometric parameters (tk,i is a vector that modifies the original geometry sk,i).”
However, Deprelle fails to teach adding ek,i to the primitive to update the set of primitives. Although Deprelle teaches value tk,i representing a “learned vector” that is added to the primitive, Deprelle does not teach or suggest that t-k,i is a sub-vector of a descriptor vector as required by claim 1.
Cohen (JP 2019521417 A, from applicant’s IDS) teaches: “the process by combiner 507 can be performed by adding residual data and corresponding points” (Pg. 5, [17]) and that “Residual ri , i = {1, 2,. . . , N} 108 can be input to transformation process 111 to generate a set of transformation coefficients 112.” (Pg. 3, [9], par. 3) However, Cohen does not create a descriptor vector including the residual, nor does Cohen discuss updating a set of primitives by adding the residual.
None of the other prior art searched or on the record teaches, suggests, or renders obvious the limitations of claim 18.
Claims 11-14 allowed.
The following is an examiner’s statement of reasons for allowance:
Claims 11 and 13 recite the limitations “determine distribution parameters comprising probabilistic parameters including a mean and variance for each primitive” and “shift and glue the set of primitives and the generated points, based on the global descriptor, using a second neural network, and based on an affinity matrix computed as pairwise inner products of primitive normal vectors.”
Li (CN 111582015 A; from Applicant’s IDS) teaches “the multi-parameter analysis platform using cloud storage: calculating the average thickness of the seal edge based on the mean value size” but does not teach calculating this mean alongside a variance for a primitive of a set of primitives for 3D shape generation.
No other references were found that explicitly teach determining a mean and variance parameter for each primitive during 3D shape generation.
Previously, He et al.: (NPL: A LINE-BASED SPECTRAL CLUSTERING METHOD FOR EFFICIENT PLANAR STRUCTURE EXTRACTION FROM LIDAR DATA) and Hotta et al.: (NPL: Statistical Analysis of Inner Products from Normal-vectors to 3D Point Cloud Clustering) were cited to teach computing an affinity matrix computed based on inner products of normal vectors. However, He and Hotta (as well as Deprelle and Groueix) do not explicitly teach shifting and gluing (or otherwise combining) primitives based on the such an affinity matrix.
Deng teaches a “feature space distance matrix” but does not calculate the matrix based on the inner products of normal vectors.
None of the other art searched or on the record teaches, suggests, or renders obvious the limitations of claims 11 and 13.
Any comments considered necessary by applicant must be submitted no later than the payment of the issue fee and, to avoid processing delays, should preferably accompany the issue fee. Such submissions should be clearly labeled “Comments on Statement of Reasons for Allowance.”
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Wikipedia (NPL: Orientation (geometry)) discusses that “orientation” may include both translation and rotation: “A rotation may not be enough to reach the current placement. It may be necessary to add an imaginary translation, called the object's location (or position, or linear position).” (Wikipedia, Pg. 1, par. 1)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VINCENT ALEXANDER PROVIDENCE whose telephone number is (571)270-5765. The examiner can normally be reached Monday-Thursday 8:30-5:00.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, King Poon can be reached on (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 published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/VINCENT ALEXANDER PROVIDENCE/Examiner, Art Unit 2617 /KING Y POON/Supervisory Patent Examiner, Art Unit 2617
Prosecution Insights