Deep Learning Image Judgment with Difference and Frequency Domain Preprocessing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image-based deep learning models face challenges in recognizing hard-to-see features and are often affected by external environments such as lighting and background noise, leading to suboptimal judgment performance.
Innovation Solution
The method involves generating difference images and converting target images into frequency domain information, which are then input into a deep learning model for judgment, enhancing feature robustness and accuracy against external factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image-based deep learning models are used for judgment, then the system is simple to operate, but the model cannot recognize hard-to-see features and is affected by external environments such as lighting and background noise
Solution Approach 1:
The patent applies preliminary action by performing image preprocessing (generating difference images and frequency domain information) before the deep learning model processes the data. This preprocessing step enhances robust features and removes noise interference in advance, allowing the model to focus on meaningful patterns during training and inference, thereby improving recognition accuracy without requiring complex modifications to the model architecture itself
Solution Approach 2:
The patent introduces intermediary representations (difference images and frequency domain information) that act as mediators between the raw input image and the deep learning model. These intermediaries transform the original image into forms that emphasize relevant features and suppress noise, enabling the model to achieve higher accuracy without directly processing the noisy original image
2Measurement precision
If preprocessing steps are added to enhance features and remove noise, then recognition performance improves, but the processing time and computational complexity increase
Solution Approach 1:
The patent extracts only the essential and robust features from the original image by generating difference images that capture local variations and frequency domain information that represents global patterns. This extraction process removes unnecessary noise and redundant information, reducing the amount of data the model needs to process while maintaining or improving judgment accuracy
Solution Approach 2:
The patent changes the parameter representation of the image data by transforming spatial domain images into frequency domain information and creating difference images. This parameter transformation converts the original pixel values into new representations that are more suitable for the deep learning model, enabling faster and more accurate processing
Data Source
AI summary
Disclosed is a system and method for judgment using deep learning model, the method for judgment using a deep learning model includes steps of outputting a predetermined judgment result when an image is input, the method comprising the steps of: receiving a target image for judgment, by a system; generating a difference image based on the received target image, wherein the difference image is an image whose pixel values are the difference values between a pixel in the target image and one of its surrounding pixels, by the system; converting the target image into frequency domain information, by the system; and inputting the difference image and the frequency domain information into the deep learning model and acquiring the judgment result output from the deep learning model, by the system.