Image Processing Apparatus Multi-Subject Certainty Weighting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing systems struggle to apply appropriate corrections to images that contain multiple photographic scenes or subjects, leading to potential adverse effects on the final correction results due to unclear scene classification.

Innovation Solution

An image processing apparatus and method that calculates certainties for various subject types within an image using feature values, weights correction values based on these certainties, and unifies them to produce a unified correction value for balanced image correction, even when images contain multiple subject types or lighting conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a single scene determination is made based on color histogram and subject detection, then the processing is simple and fast, but the correction accuracy deteriorates when multiple subject types are present

Engineering Contradiction:
Improvescene classification accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the scene determination process by calculating certainties for multiple subject types (snapshot, portrait, commemorative shot, scenery, sunset, night scene, etc.) separately and independently. Each subject type is evaluated with its own certainty value, allowing the system to handle multiple subjects without increasing overall processing complexity. This segmentation enables accurate multi-subject classification while maintaining simple individual subject detection algorithms.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces certainty values as a new parameter to quantify the likelihood of each subject type. By changing from binary classification to probabilistic classification with certainty values, the system can accurately represent images containing multiple subject types. The correction process then uses these certainty parameters to weight and combine correction values from different subject types, achieving accurate correction without complex processing.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If correction processing is applied based on a single determined scene, then the processing speed is fast, but the correction appropriateness deteriorates for images with multiple subject types

Engineering Contradiction:
Improveprocessing speedVSAvoidcorrection appropriateness
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary calculation of certainty values for all possible subject types before applying correction. By pre-calculating these certainties and storing them, the system avoids complex real-time calculations during the correction phase. This preliminary action maintains fast processing speed while ensuring that all necessary information for accurate multi-subject correction is available beforehand.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent maintains continuous correction capability across all subject types by calculating correction values for each subject type and then combining them using their respective certainty values. This continuous approach ensures that correction is always appropriate regardless of how many subject types are present in the image, while maintaining efficient processing through the use of pre-calculated values and simple weighted combination operations.

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentUS7894673B2Image processing apparatus and method, and image processing computer readable medium for processing based on subject type
Publication Date: 2011.02.22 FUJIFILM CORP
  • US7894673B2 patent drawing
  • US7894673B2 patent drawing
  • US7894673B2 patent drawing

AI summary

A certainty calculating circuit calculates certainties representing to what degrees of confidence respective ones of images representing a predetermined plurality of types of subject are contained in an image represented by accepted image data, the calculation being performed for every prescribed type of subject based upon a feature value obtained from the accepted image data. A density correction value calculating circuit calculates density correction values with regard to respective ones of the plurality of subject types. Upon being weighted by respective ones of weights that are based upon the calculated certainties of each of the subject types, the plurality of density correction values are unified in a density correction value unifying circuit. The density of the image represented by the accepted image data is corrected in an image correcting circuit using the unified density correction value.