FPGA Image Presegmentation for Real-Time Contrast Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional camera systems face high computing and processing demands due to slow pixel access times when performing image presegmentation into regions of interest and no interest, particularly in real-time applications like code reading, where only specific image regions are relevant.

Innovation Solution

The system divides images into fixed-size part regions and calculates contrast values in two steps, with the FPGA handling initial pixel summations and the microprocessor performing subsequent calculations, reducing the need for pixel access and enabling faster presegmentation by distributing workload between preprocessing and structure recognition units.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If conventional CPU-based algorithms are used for image presegmentation, then flexibility in implementing different image processing processes is improved, but processing speed deteriorates due to slow pixel access times

Engineering Contradiction:

Improveflexibility in implementing image processing processesVSAvoidprocessing speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent divides the image processing task into two segments: an FPGA-based preprocessing unit that handles pixel access and contrast calculation for part regions, and a microprocessor-based structure recognition unit that performs higher-level analysis. This segmentation allows each component to operate at optimal speed and complexity levels, resolving the contradiction between flexibility and processing speed.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary data structure (summed values) that is calculated by the FPGA and passed to the microprocessor. This intermediary reduces the data transfer burden and enables the microprocessor to focus on complex structure recognition without being bottlenecked by slow pixel access, thus maintaining both speed and flexibility.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If all pixels are accessed to calculate contrast values for presegmentation, then measurement precision of image regions is improved, but processing time increases due to numerous pixel accesses

Engineering Contradiction:

Improveaccuracy of contrast value calculationVSAvoidpresegmentation processing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary calculation of summed values for part regions in the FPGA before the microprocessor needs to analyze contrast. This preliminary action pre-computes the necessary data, reducing the time the microprocessor would otherwise spend accessing and processing individual pixels, thus maintaining precision while reducing processing time.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical sequential pixel access process with a parallel FPGA-based computation system that calculates summed values simultaneously for multiple part regions. This substitution dramatically reduces processing time while maintaining the accuracy needed for contrast-based presegmentation.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Manufacturing precision

If high-resolution images are processed, then image quality is improved, but computing capacity requirements increase due to large amounts of image data

Engineering Contradiction:

Improveimage resolution qualityVSAvoidcomputing capacity requirements

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent extracts and processes only the necessary contrast information for part regions using FPGA hardware, separating this computationally intensive task from the main microprocessor workload. This extraction allows high-resolution images to be processed by utilizing the FPGA's parallel processing capability for initial analysis, reducing the computing capacity burden on the microprocessor while maintaining image quality.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS9008426B2Generating an image presegmented into regions of interest and regions of no interest

Publication Date: 2015.04.14 SICK AG

AI summary

A camera system (10) is provided for generating an image presegmented into regions (106a-b) of interest and of no interest, having an evaluation unit (20) which is designed to divide the raw image into part regions (106a-b) to calculate a contrast value for each part region (106a-b) and to decide with reference to the contrast value whether the respective part region (106a-b) is a region of interest (106a) or a region of no interest (106b). In this respect, the evaluation unit (20) has a preprocessing unit (22) which is implemented on an FPGA, which respectively accesses the pixels of a part region (106a-b) and generates summed values (a), b) for the respective part region (106a-b) and has a structure recognition unit (24) which calculates the contrast value of the part region (106a-b) from its summed values (a, b) without accessing pixels of the part region (106a-b).