Image Processing Device Using Precomputed Execution Results
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Combining image recognition technology with pixel color-based methods for identifying areas in images increases processing time significantly, compromising efficiency while maintaining precision.
Innovation Solution
An image processing device and method that utilizes preprocessing results, such as position and color data, to be held and applied sequentially to subsequent images, reducing the need for repeated high-precision processing and minimizing overall processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image recognition technology is combined with pixel color-based methods for identifying areas in images, then measurement precision is improved, but processing time increases significantly
Solution Approach 1:
The patent applies preliminary action by performing image recognition technology (high-precision method) in advance to generate execution results, which are then stored and reused for subsequent pixel color-based processing. This allows the system to maintain high precision through the initial comprehensive analysis while avoiding repeated execution of the time-consuming recognition algorithm, thus resolving the contradiction between precision and processing time.
Solution Approach 2:
The patent changes the parameter of method selection dynamically. Instead of consistently using the slower but more precise image recognition technology, the system switches to using pre-computed execution results from the recognition technology combined with faster pixel color-based methods for subsequent processing. This parameter change allows the system to achieve both high precision (through the initial recognition) and fast processing (through reused results).
2Measurement precision
If high-precision image processing is performed on every image, then measurement precision is improved, but productivity decreases due to long processing time
Solution Approach 1:
The system performs high-precision image recognition as a preliminary action once, generates execution results, and stores them for reuse. This preliminary comprehensive analysis ensures high precision is maintained, while the reused results eliminate the need to repeat the time-consuming recognition process on every subsequent image, thereby maintaining high productivity.
Solution Approach 2:
The patent uses copying by storing the execution results of the high-precision image recognition technology and reusing these copied results for subsequent processing. Instead of re-executing the full high-precision recognition algorithm on every image, the system copies and applies the pre-computed results, which maintains precision while dramatically improving processing speed and productivity.
Data Source
AI summary
Provided is an image processing device, an image processing method, and a program which can prevent a total processing time from significantly increasing while maintaining precision of image processing at a high level, where an image acquiring section sequentially acquires images generated by imaging a predetermined subject to be imaged; an image processing executing section executes, in each of sequentially-arriving processing periods, image processing on the image acquired by the image acquiring section; a preprocessing execution result output section outputs an execution result of preprocessing performed on the image in part of the sequentially-arriving processing periods, the image having been acquired by the image acquiring section before the part of the sequentially-arriving processing periods; an execution result holding section keeps holding the execution result output by the preprocessing execution result output section at least until the execution result is output next time by the preprocessing execution result output section; and the image processing executing section executes the image processing by applying the execution result held in the execution result holding section to the image acquired by the image acquiring section.