Image Processing Device for Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies fail to effectively eliminate light information from images, leading to deteriorated object recognition rates when correcting gradation in scenes with large light and dark differences, such as backlit scenes, as they often retain illumination information even after correction.
Innovation Solution
An image processing device and method that corrects pixel levels based on low-frequency components from peripheral images, using threshold values to determine whether to apply corrections, and performs blending processes to eliminate illumination components, thereby enhancing object recognition by removing unnecessary light information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If gradation correction is performed in scenes with large light and dark differences, then image quality is improved, but object recognition rate deteriorates due to retained illumination information
Solution Approach 1:
The image is divided into frequency components (low-frequency illumination component and high-frequency object component). By separating these components, the patent applies different processing to each: the illumination component is corrected for gradation while the object component is preserved for recognition, thus resolving the contradiction between image quality improvement and recognition rate maintenance.
Solution Approach 2:
The patent extracts the illumination component from the original image using low-pass filtering. This extracted illumination component is then used for gradation correction while being excluded from the final recognition image, effectively removing the harmful illumination information that would otherwise deteriorate recognition rate.
2Manufacturing precision
If illumination information is retained in corrected images, then gradation correction effectiveness is improved, but object recognition performance deteriorates
Solution Approach 1:
The patent applies different quality requirements to different parts of the image processing pipeline. The illumination component is processed with gradation correction to improve visual quality, while the object component is preserved with minimal processing to maintain recognition performance. This local differentiation resolves the contradiction between correction effectiveness and recognition performance.
Data Source
AI summary
An image that enables proper object recognition is generated. An image processing device includes an image processing unit that processes an image. The image processing unit included in this image processing device corrects the level of a pixel constituting the image. That is, the image processing unit corrects the level of a target pixel constituting the image on the basis of a low-frequency component in a peripheral image of a pixel constituting the target image. Here, the peripheral image is made up of a plurality of pixels that includes the target pixel and is located around the target pixel.