Depth and Intensity MSER Fusion for Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feature detection methods, such as MSER, are highly dependent on image intensity and struggle with reliability in identifying and tracking objects under varying conditions, leading to issues with extraneous region identification.
Innovation Solution
The method processes both depth and intensity data to identify 'strong' maximally stable extremal regions (MSERs) that are consistent across different viewing conditions, reducing extraneous region identification by leveraging both depth and intensity MSERs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If MSER is used for feature detection based on image intensity, then feature detection capability is improved, but reliability under varying conditions deteriorates due to high dependence on intensity
Solution Approach 1:
The patent combines intensity-based MSER detection with depth-based MSER detection to create a unified feature detection system. By merging the strengths of both intensity and depth information, the system achieves improved reliability under varying conditions while maintaining feature detection capability. The combination allows the system to overcome the limitations of intensity-only methods by incorporating depth data that is less sensitive to lighting variations.
Solution Approach 2:
The patent introduces depth data as an intermediary element that mediates between the intensity information and the final feature detection results. By using depth MSERs as an additional source of information, the system can validate and confirm features detected from intensity data, thereby improving reliability without sacrificing detection capability.
2Quantity of substance
If intensity-based MSERs are used for object identification, then number of identified regions is increased, but reliability deteriorates due to identification of extraneous regions
Solution Approach 1:
The patent extracts and isolates depth-based MSERs as a separate verification layer. By taking out the depth information and using it to filter the intensity-based MSER results, the system removes extraneous regions that were incorrectly identified. This extraction process allows the system to maintain the quantity of identified regions from intensity data while eliminating the unreliable ones through depth verification.
Solution Approach 2:
The patent implements a feedback mechanism where depth MSER detection results are used to validate and refine the intensity MSER results. The depth information provides feedback that confirms or rejects potential object regions, thereby improving the reliability of object identification while preserving the comprehensive region detection capability of the intensity-based approach.
3Reliability
If only depth data is used for object detection, then reliability is improved by reducing extraneous regions, but detection completeness deteriorates
Solution Approach 1:
The patent merges depth-based MSER detection with intensity-based MSER detection to achieve both reliability and completeness. By combining the two detection approaches, the system leverages the reliability advantage of depth data while compensating for its lower completeness through the addition of intensity data, which can detect additional regions that depth alone might miss.
Data Source
AI summary
Methods and systems for detecting and/or tracking one or more objects utilize depth data. An example method of detecting one or more objects in image data includes receiving depth image data corresponding to a depth image view point relative to the one or more objects. A series of binary threshold depth images are formed from the depth image data. Each of the binary threshold depth images is based on a respective depth. One or more depth extremal regions in which image pixels have the same value are identified for each of the binary depth threshold images. One or more depth maximally stable extremal regions are selected from the identified depth extremal regions based on change in area of the one or more respective depth extremal regions for different depths.