Image Stack Data Structure Synchronization Compensation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image recording systems face synchronization issues when capturing moving objects, leading to transport artifacts and distortions, especially at high magnifications, which impede visual quality control and accurate three-dimensional reconstruction.

Innovation Solution

A method for creating an image stack data structure that compensates for synchronization problems by aligning sensor rows perpendicular to the transport direction, calculating subpixel distances between entries, and adjusting recording indices using an optimization process to ensure even distribution and precise brightness value reproduction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high-precision linear drives or motion sensors are used to prevent synchronization problems, then synchronization accuracy is improved, but device complexity and cost increase

Engineering Contradiction:
Improvesynchronization accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces mechanical synchronization systems (high-precision linear drives or motion sensors) with a computational approach. By arranging sensor rows perpendicular to the transport direction and using post-processing algorithms to calculate subpixel distances and adjust recording indices, the system achieves synchronization compensation without complex mechanical components. This substitutes mechanical precision requirements with computational processing.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If transport speed is increased to improve productivity, then recording speed is improved, but transport artifacts and distortions worsen

Engineering Contradiction:
Improverecording speedVSAvoidimage accuracy
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent performs preliminary arrangement of sensor rows perpendicular to the transport direction before recording begins. This spatial configuration is pre-established to enable subsequent computational compensation. By preparing the sensor geometry in advance, the system can handle high-speed transport without requiring real-time mechanical adjustments, thus maintaining both speed and accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements a feedback mechanism through computational processing of recorded data. By calculating subpixel distances between corresponding entries in the image stack and adjusting recording indices based on these calculations, the system compensates for transport artifacts after data collection. This feedback loop corrects distortions that occur during high-speed transport.

Inventive Principle:
Principle #23Feedback

3Device complexity

If sensor rows are arranged parallel to transport direction, then device simplicity is improved, but transport artifact compensation capability deteriorates

Engineering Contradiction:
Improvesensor arrangement simplicityVSAvoidartifact compensation capability
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent employs an asymmetric sensor arrangement where sensor rows are positioned perpendicular to the transport direction rather than parallel. This asymmetric configuration relative to the transport axis enables the system to capture spatial information necessary for computing transport artifacts. The perpendicular arrangement creates a geometric relationship that facilitates the calculation of subpixel distances and subsequent compensation.

Inventive Principle:
Principle #4Asymmetry

Data Source

PatentEP3518180B1Method for producing an image stack data structure
Publication Date: 2020.09.16 AIT AUSTRIAN INSTITUTE OF TECNOLOGY GMBH
  • EP3518180B1 patent drawingFigure 1~2
  • EP3518180B1 patent drawingFigure 3a~3b
  • EP3518180B1 patent drawingFigure 4

AI summary

The invention relates to a method for creating an image stack data structure (S), wherein: - an object (1) is moved along a transport direction (x) through the recording area of ​​an image acquisition unit (2); - images (v1, ..., v4) are created by the image acquisition unit (2); - the image acquisition unit (2) has a number of sensor rows (21a, 21b, 21c) that are arranged one behind the other, in particular parallel to each other, normal to the transport direction (x) of the object (1) and in relation to the transport direction (x), in particular forming rows of an area sensor; - the individual images (v1, ..., v4) each comprise row images, each of which is created by means of a sensor row (21a, 21b, 21c) at predetermined recording times; - the predetermined recording times are the same for all images and correspond to the transport progress along the transport direction (x).- a three-dimensional image stack data structure (S) is created from the images (v1, ..., v4) of the object (1), wherein the individual entries (E1, E2, E3) of the image stack data structure (S) contain, for a number of image indices (t1, ..., t9), at least one brightness value for the individual pixels of the sensor rows (21a, 21b, 21c), wherein the indices of the three-dimensional image stack data structure (S) are defined by i) an integer image index (t1, ..., t9) characterizing the transport progress to which the relevant brightness value was created, ii) a row index (z1, ..., z4) characterizing the sensor row (21a, 21b, 21c), in particular its position in the transport direction (x) in which the pixel that determined the relevant brightness value is located, and iii) the column position (y1, ..., y5) of the pixel that has determined the relevant brightness value with respect to the sensor rows (21a, 21b,21c) normal to the transport direction (x), - wherein for individual selected entries, in particular for all entries (E1, E2, E3), of the image stack data structure (S) - starting from the respective selected entry (E2; E5), which is assigned to a sensor row (21a, 21b, 21c), a search is performed for at least two entries whose brightness values ​​were created by other sensor rows (21a, 21b, 21c) and which are assigned to the same column position (y1, ..., y5), which show the same object area as the respective selected entry (E2; E5), and whose row indices (z1, z2, z3) are different from the row index of the selected entry, wherein in particular the difference of their row indices to the row index of the selected entry (E2; E5) has one of several predefined values, - the distances (d21, d23) with respect to the acquisition index (t1, t5, t9) are each between the selected entry (E1, E2,E3) and each of the entries thus found are calculated using subpixel arithmetic, - a distance mean value (d21*, d23*) of the distances (d21, d23) for individual image indices (t1, t5, t9) and row indices (z1, .., z4) across selected, in particular all, column positions (y1, ..., y5), in particular for both distances (d21, d23) separately, is formed, - for a number of pairs comprising one of the image indices (t1, t5, t9) and one of the row indices (z1, .., z4) each, to which the respective distance mean values ​​(di°, di∼; d21*, d23*) are assigned, and - to a number of selected elements (E2*, E5*) further elements (E1*, E3*, E4,1*, E4,2*, E6,1*, E6,2*) are assigned a row index relative to the respective selected element (E2*, E5*), where the difference between the unadjusted recording indices (t1, ..., t9) of the selected element (E2*,E5*) and the further element (E1*, E3*, E4,1*, E4,2*, E6,1*, E6,2*) corresponds to, or closely approximates, the respective distance mean value (d21*, d23*) assigned to the selected element (E2*, E5*), where, if necessary, the two further elements (E4,1*, E4,2*, E6,1*, E6,2*) with the same row index are jointly assigned to the respective selected element (E2*, E5*) with the two differences of the unadjusted acquisition indices that are closest to the respective distance mean value (d21*, d23*), and - an index adjustment of the original acquisition index (t1, t5, t9) is performed and, for each acquisition index (t1, t5, t9), a subpixel-accurate adjusted acquisition index (t'1, t'5, t'9) is used as a replacement for the original acquisition index. (t1, t5, t9) is determined by means of an optimization procedure (O) such that the following constraints are satisfied: - the individual differences between the adjusted intake index (t'1,...) of a selected element (E2*) and the adjusted intake index (t'1, ...) of an element (E1*) assigned to it are equal to each other, - wherein the distance mean values ​​(d21*, d23*) are optionally weighted with a compensation factor (k21*, k23*) inversely proportional to the row index difference of the entries or elements used for its determination, and wherein in particular the adjusted intake index (t'1, ...) of several jointly assigned elements (E4,1*, E4,2*) is determined by weighting the adjusted intake indices (t'1, ...) of the jointly assigned elements (E4,1*, E4,2*), - the adjusted intake index (t'1, ...) of an element (E1*, ...) is as close as possible to the respective unadjusted intake index (t1, ...) of the element in question (E1*, ...).,