Image Processing Apparatus Parallel Evaluation Area Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing technologies face inefficiencies and increased costs due to the need for multiple signal processing units to calculate evaluation values for detection areas that span multiple divided areas, leading to increased processing time and cost.
Innovation Solution
An image processing apparatus and method that sets evaluation areas to entirely fit within individual divided image areas, avoiding the calculation of evaluation values for areas that span multiple areas, thereby optimizing processing efficiency and reducing costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple detection areas are set to evaluate the whole video area, then the evaluation value more appropriately reflects the overall condition, but detection areas may span multiple divided areas requiring complex inter-unit cooperation and data transfer
Solution Approach 1:
The video area is divided into multiple divided areas that are processed in parallel by multiple signal processing units. Each unit independently calculates evaluation values for detection areas contained within its assigned divided area, eliminating the need for inter-unit data transfer and complex cooperation mechanisms.
Solution Approach 2:
Each signal processing unit is assigned specific divided areas and independently processes detection areas within those boundaries. This local processing approach allows each unit to operate autonomously with its own data, improving efficiency while maintaining overall evaluation accuracy through the aggregation of local results.
2Measurement precision
If detection areas span multiple divided areas, then comprehensive evaluation is achieved, but data transfer between signal processing units and additional memory are required increasing cost
Solution Approach 1:
The system segments both the video area and detection areas such that each detection area is contained within a single divided area. This segmentation strategy eliminates the need for data transfer between signal processing units, as each unit processes its assigned detection areas independently using only its local data.
3Measurement precision
If multiple signal processing units cooperate to calculate evaluation values for spanning detection areas, then complete evaluation is possible, but processing time increases
Solution Approach 1:
The video and detection areas are segmented so that each detection area falls entirely within one divided area. This allows multiple signal processing units to calculate evaluation values in completely parallel fashion without any cooperation or synchronization overhead, dramatically reducing processing time while maintaining comprehensive evaluation coverage.
Solution Approach 2:
The system pre-divides the video area into multiple divided areas and pre-assigns detection areas to specific divided areas before processing begins. This preliminary organization ensures that during the actual evaluation phase, each signal processing unit can immediately begin independent parallel calculation without any inter-unit coordination requirements.
Data Source
AI summary
There is provided an image processing apparatus comprising: an acquisition unit configured to acquire an image signal; a setting unit configured to set, in an image expressed by the image signal, a plurality of evaluation areas to be targets of evaluation value calculation; a dividing unit configured to divide the image expressed by the image signal into a plurality of divided image areas; and a plurality of calculation units configured to acquire, from the dividing unit, divided image signals respectively corresponding to the plurality of divided image areas, and calculate evaluation values for the plurality of evaluation areas based on evaluation area image signals respectively corresponding to the plurality of evaluation areas included in the divided image areas. Among the plurality of evaluation areas, the plurality of calculation units do not calculate an evaluation value for each evaluation area that spans two or more of the divided image areas.