Product Image Processor Scattering Filter Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Product images displayed as thumbnails are too small, making it difficult to differentiate between similar products like nail polish and mayonnaise, where colors and label descriptions are not easily distinguishable.
Innovation Solution
A product image processor that calculates the degree of scattering of pixel values across common compositions in product images, generates a filter based on this scattering, and applies it to extract feature images that highlight product features by maximizing the sum of scattering in rectangular regions, using a monotone increasing function and affine transformation to standardize image orientation and size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If product images are displayed as thumbnails to show multiple products, then the quantity of products displayed increases, but the visibility of product features deteriorates
Solution Approach 1:
The patent divides the product image into multiple regions based on scattering degree, extracting only the feature-containing regions (product body, label, cap) while excluding background regions. This segmentation allows multiple products to be displayed as thumbnails while maintaining feature visibility by showing only the relevant segmented portions of each product.
Solution Approach 2:
The patent applies different processing to different regions of the image based on their scattering characteristics. Regions with high scattering degree (product features) are extracted and enhanced, while regions with low scattering degree (background) are excluded. This local quality approach ensures that thumbnail displays maintain feature visibility by selectively preserving only the important local regions.
2Stability of the object's composition
If common composition is used for product images to maintain brand consistency, then the uniformity of product presentation improves, but the ability to highlight individual product features deteriorates
Solution Approach 1:
The patent extracts the feature-containing regions from the common composition by calculating scattering degree and identifying regions with high scattering values. This extraction process separates the individual product features (label text, product color, cap details) from the common background composition, allowing brand consistency to be maintained while individual features are highlighted for differentiation.
Solution Approach 2:
The patent applies region-specific processing based on scattering characteristics, preserving and enhancing only the local regions that contain product-specific features while maintaining the common composition structure. This allows the overall brand image consistency to be preserved while individual product features are differentiated through selective enhancement of high-scattering regions.
3Loss of information
If image processing is performed on the entire product image, then the completeness of information is maintained, but the processing time increases
Solution Approach 1:
The patent segments the image processing task by first calculating scattering degree for all pixels, then identifying and extracting only the regions with high scattering degree (product features). This segmentation allows the system to process only the relevant feature regions rather than the entire image, reducing processing time while maintaining completeness of product information through selective extraction.
Solution Approach 2:
The patent extracts only the essential feature-containing regions from the full product image by using scattering degree as a criterion. This extraction removes unnecessary background and non-feature areas, allowing the processing to focus only on the important product information, thereby reducing processing time while maintaining information completeness.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
In order to obtain feature images representing features of products from product images, in a product image processor (101), a receiver (102) receives product images each representing one product in a group of products with a common composition; a calculator (103) calculates a degree of scattering of pixel values at each of the positions in the composition from a pixel value at each of the positions in each of the received product images; a generator (104) generates a filter with a degree of transmittance defined at each of positions in the composition on the basis of a degree of scattering calculated at each of the positions; and an applicator (105) applies the generated filter to each of the received product images, thereby obtaining feature images, each representing a feature of each of the products.