Video Object Range Estimation Using Iterative Feature Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video surveillance technologies face limitations in accurately estimating the range of objects in video images, particularly for stationary objects and those with unknown sizes or oblique motion, leading to unreliable classifications, as existing methods like passive ranging are less reliable and expensive solutions like laser ranging and stereo imaging require significant resources and labor for implementation.
Innovation Solution
A method that segments digital images into regions, processes them using sensed-feature-based classification to identify prominent objects, estimates their range, and iteratively converts sensed features to physical features through a physical-feature-based classification, improving accuracy and reliability by re-running the processes until a threshold is met, allowing for continuous refinement of object classification and range estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If passive ranging is used to estimate object range, then implementation cost is reduced, but measurement precision and reliability deteriorate
Solution Approach 1:
The system performs preliminary classification of objects into prominent and non-prominent categories before range estimation. By pre-sorting objects based on classification reliability and prominence thresholds, the system can apply more accurate range estimation methods only to the most relevant objects, improving overall measurement precision without requiring expensive laser ranging for all objects
Solution Approach 2:
The patent segments objects into different categories (prominent vs. non-prominent) based on their classification reliability and visual characteristics. This segmentation allows the system to apply different range estimation strategies to different object types, using simpler methods for less critical objects while reserving more accurate methods for prominent objects, thus balancing cost and precision
2Measurement precision
If laser ranging or stereo imaging is used to improve range estimation accuracy, then measurement precision improves, but device complexity and cost increase
Solution Approach 1:
The patent extracts and utilizes existing video image data and classification information that are already available in the surveillance system. By taking out and reusing this existing data for range estimation, the system avoids adding complex external ranging devices while still achieving improved measurement precision through intelligent data processing
Solution Approach 2:
The patent replaces mechanical/optical ranging systems (laser range finders, stereo cameras) with a computational approach that uses image processing, object classification, and iterative algorithms. This substitution eliminates the need for additional hardware while achieving comparable or superior range estimation accuracy through software-based methods
3Reliability
If iterative classification and range estimation processes are applied, then reliability improves, but processing time increases
Solution Approach 1:
The patent applies partial iterative processing by stopping the classification and range estimation loops when predefined reliability thresholds are met. Rather than continuously processing until perfect accuracy is achieved, the system performs enough iterations to reach sufficient reliability, then halts to avoid unnecessary time consumption. This partial action approach balances reliability improvement with processing time constraints
4Measurement precision
If all objects are processed with high accuracy methods, then measurement precision improves, but productivity decreases
Solution Approach 1:
The patent segments objects into prominent and non-prominent categories based on their classification reliability and visual characteristics. Prominent objects receive high-accuracy processing, while non-prominent objects are processed with simpler, faster methods. This segmentation strategy maintains high productivity by avoiding expensive processing on all objects while ensuring accurate measurement precision for the most important objects
Data Source
AI summary
An image is processed by a sensed-feature-based classifier to generate a list of objects assigned to classes. The most prominent objects (those objects whose classification is most likely reliable) are selected for range estimation and interpolation. Based on the range estimation and interpolation, the sensed features are converted to physical features for each object. Next, that subset of objects is then run through a physical-feature-based classifier that re-classifies the objects. Next, the objects and their range estimates are re-run through the processes of range estimation and interpolation, sensed-feature-to-physical-feature conversion, and physical-feature-based classification iteratively to continuously increase the reliability of the classification as well as the range estimation. The iterations are halted when the reliability reaches a predetermined confidence threshold. In a preferred embodiment, a next subset of objects having the next highest prominence in the same image is selected and the entire iterative process is repeated. This set of iterations will include evaluation of both of the first and second subsets of objects. The process can be repeated until all objects have been classified.