3D Data Decoding via Projection Direction and Patch Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for decoding 3D data encoded using a video-based approach are restricted by decoding order and range, requiring the entire dataset to be decoded to access a desired portion, leading to complex and inefficient processing.
Innovation Solution
An image processing apparatus and method that generates and decodes bit streams including projection direction information, allowing for flexible projection directions and patch placement, enabling partial decoding and reducing processing load by associating projection directions with identifiers and encoding units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire decoding target is decoded to access a desired portion, then the desired portion can be obtained, but the processing complexity and time increase significantly
Solution Approach 1:
The patent divides the 3D data into multiple patches that can be independently decoded. Each patch corresponds to a specific region in the decoded image, allowing the decoder to selectively decode only the patches containing the desired portion rather than the entire decoding target. This segmentation enables partial decoding and reduces unnecessary processing time.
Solution Approach 2:
The patent implements partial decoding by allowing the decoder to decode only a subset of patches rather than the complete set. The bitstream includes information about which patches are available and their spatial relationships, enabling the decoder to perform just enough decoding to access the desired portion without excessive processing of unrelated data.
2Ease of manufacture
If traditional video-based encoding is used for 3D data, then encoding can be performed, but decoding order and range are restricted
Solution Approach 1:
The patent organizes patches in a two-dimensional array structure with row and column indices, adding spatial dimension information to the encoding. This dimensional organization allows the decoder to navigate and access patches flexibly based on their spatial positions, overcoming the linear decoding order restrictions of traditional video-based encoding methods.
Solution Approach 2:
The patent creates a universal patch-based encoding framework that can handle various types of 3D data and support multiple decoding scenarios. The same encoding structure enables both full decoding and partial decoding, sequential and random access, making the system versatile for different application requirements while maintaining encoding capability.
3Quantity of substance
If patches are densely packed in encoding units, then data compactness is improved, but occlusion and patch decomposition increase
Solution Approach 1:
The patent applies different packing strategies to different regions based on their characteristics. Patches are organized in a structured two-dimensional array where spatial relationships are preserved, allowing regions with potential occlusion issues to be handled differently from non-occluded regions. This local quality approach maintains data compactness while reducing harmful occlusion effects in critical areas.
Solution Approach 2:
The patent performs preliminary organization of patches into a structured array before encoding, establishing spatial relationships and identifying potential occlusion issues in advance. This preliminary action allows the encoder to prepare appropriate handling strategies for each patch, reducing decomposition problems during decoding while maintaining compact representation.
Data Source
AI summary
An image processing apparatus and method that make it possible to decode encoded data of 3D data with increased ease. A bit stream is generated which includes projection direction information including information relating to a projection direction of position information of 3D data representative of a three-dimensional structure on a two-dimensional plane and encoded data of a geometry image obtained by projecting the position information on the two-dimensional plane.