Light Field Distance Determination Using Intermediate Image Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the distance of an imaged object using light field recording devices are resource-intensive, especially when dealing with a large number of intermediate images, and require complex matching processes.
Innovation Solution
A method that evaluates the quality of intermediate images by determining the highest quality image for each pixel position, interpolating based on similar object distances, and using a combination of sharpness and agreement measures to assign object distances, allowing for resource-efficient processing and precise distance determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex matching methods are used to determine distance from light field data, then measurement precision is improved, but device complexity and resource requirements increase
Solution Approach 1:
The patent segments the light field data into multiple intermediate images corresponding to different focal planes. By dividing the complex light field analysis into discrete depth layers, the method simplifies the matching process while maintaining precision. Each intermediate image can be processed independently, reducing overall computational complexity.
Solution Approach 2:
The patent introduces intermediate images as a mediator between the raw light field data and the final distance determination. These intermediate images represent different focal planes and serve as simplified representations that facilitate easier comparison and matching, reducing the complexity of direct light field analysis.
2Measurement precision
If a large number of intermediate images are created for accurate distance determination, then measurement precision is improved, but resource requirements increase
Solution Approach 1:
The patent applies partial action by selecting only the necessary number of intermediate images for processing. Instead of evaluating all possible focal planes, the method identifies and processes only those intermediate images that contain relevant information for the specific distance determination task, reducing resource consumption while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary filtering and selection of intermediate images before full processing. By pre-identifying which intermediate images are most relevant based on initial analysis, the method avoids unnecessary processing of redundant images, thereby reducing computational resources while preserving measurement precision.
3Measurement precision
If all superimposed intermediate images are fully evaluated, then measurement precision is improved, but productivity decreases due to high resource load
Solution Approach 1:
The patent extracts only the essential information from the superimposed intermediate images. By identifying and extracting key features and relevant data points rather than processing complete image evaluations, the method maintains measurement precision while significantly reducing the computational burden and improving processing efficiency.
Solution Approach 2:
The patent applies partial evaluation by processing only the critical portions of intermediate images that contribute most to distance determination. This selective approach avoids the resource-intensive full evaluation of all image data while preserving the accuracy needed for precise measurements.
Data Source
Figure 1~2
Figure 3~6
Figure 4~5
AI summary
Method for determining the distance of an object (2), wherein the light field is determined using a light field recording device (1) and provided as an image stack (L), wherein the image stack (L) comprises sub-images (L1, ..., Lm) and each sub-image (L1, ..., Lm) is characterized by a different recording point, a quality measure (M) of a pixel is specified, which is determined on the basis of the intensity values of the surroundings, wherein the quality measure (M) comprises two sub-steps (M1, M2), namely a linear first sub-step (M1) in which an intermediate result is determined using the intensity values of the surrounding pixels, and at least one non-linear second sub-step (M2) in which a quality value is determined from the intermediate result by means of a non-linear operation. For each of the sub-images (L1, ..., Lm), an intermediate image (Z1, ..., Zm) is created. For the individual pixels of each sub-image (L1, ...For each object distance (Lm), the intermediate result is determined and assigned to the intermediate image (Z1, ..., Zm). For each object distance (d1, ..., dn), a displacement vector (V1, ..., Vm) is determined. The intermediate image (Z1, ..., Zm) assigned to the sub-image (L1, ..., Lm) is shifted according to the determined displacement vector (V), and a superimposed intermediate image (F1, ..., Fn) of the intermediate images (Z1, ..., Zn) is created. The object distance (d1, ..., dn) is assigned to this superimposed intermediate image. In the pixels of the superimposed intermediate images (F1, ..., Fn), the nonlinear operation of the quality measure (M) is applied separately, pixel by pixel. For each superimposed intermediate image (F1, ..., Fn), a quality image (S) is created. For each pixel position, a quality image (S) is sought for which the highest quality value was determined. The object distance (d1, ..., d1) assigned to this one quality image is used to determine the distance.