Selective Image Compression via ROI Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing and compression techniques for machine-readable symbols face challenges in efficiently transmitting high-resolution images due to information loss from methods like JPEG conversion and image cropping, which can remove embedded data and binarized information essential for OCR and camera operation details.
Innovation Solution
The method involves identifying regions of interest (ROIs) in images and averaging non-ROI pixel values into a single value, allowing for lossless compression by creating a bit-wise map indicating ROI and non-ROI regions, thus preserving high-resolution ROIs while significantly reducing the size of the image for transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If image compression techniques like JPEG conversion are used to reduce image size for transmission, then bandwidth consumption is reduced, but information loss occurs including embedded data and binarized data
Solution Approach 1:
The image is segmented into multiple blocks (e.g., 16x16 pixel blocks), and each block is independently evaluated to determine if it contains a machine-readable symbol. This allows selective processing where only symbol-containing blocks are preserved at full resolution while other blocks are compressed or averaged, thus reducing overall bandwidth consumption without losing critical information.
Solution Approach 2:
Different quality levels are applied to different regions of the image based on their content. Regions containing machine-readable symbols maintain full resolution and quality, while regions without symbols use lower quality representation through averaging or compression. This local quality differentiation ensures critical data is preserved while reducing overall bandwidth requirements.
2Quantity of substance
If image cropping is used to remove unnecessary parts of the image, then image size is reduced, but embedded data and important information may be removed
Solution Approach 1:
Instead of cropping the entire image, the method segments it into blocks and evaluates each block individually. This granular approach ensures that even if parts of the image are compressed or averaged, the blocks containing machine-readable symbols and their associated embedded data are identified and preserved at full quality.
Solution Approach 2:
The system performs preliminary analysis of each image block to identify the presence of machine-readable symbols and embedded data before applying compression. This preliminary detection action ensures that critical information locations are known in advance, allowing the compression algorithm to avoid or minimize loss in those specific regions.
3Measurement precision
If high-resolution images are transmitted in real-time, then image quality is maintained, but transmission difficulty increases due to large data size
Solution Approach 1:
The high-resolution image is divided into blocks, allowing selective transmission of full-resolution data only for blocks containing machine-readable symbols. Other blocks can be transmitted at lower resolution or even skipped entirely, dramatically improving transmission efficiency while maintaining sufficient quality for the critical symbol-containing regions.
Solution Approach 2:
Instead of transmitting the entire high-resolution image, the system applies partial action by transmitting full-resolution data only for the necessary portions (symbol-containing blocks). This partial transmission approach achieves the minimum required quality for successful symbol recognition while significantly reducing bandwidth consumption and improving transmission speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods of operation for an image processor system to process images to locate two-dimensional regions which are likely to contain machine-readable symbol data or text. Such regions of interest (ROIs) may be preserved at full resolution, whilst the contents of non-ROIs are averaged into a single pixel value. Transition densities in an image may be converted into a numeric value. Such transition densities may be indicative of the presence of data of interest, such as textual data and/or machine-readable symbol data. The pixels values for the pixels in the ROIs may be sent to a decoder unchanged (i.e., full resolution), which absolves the decoder from having to perform any ROI location computations. Modified or altered images may be compressed to much smaller size files while maintaining lossless ROIs, which allows for transmission of such images to processor-based devices over a data communications channel in real time.