Neural Network Image Processing for Appearance Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing devices require users to determine combinations or sequences of image processing through trial and error, resulting in a significant burden, as there is no efficient method to obtain a desired output image from an input image for appearance inspection.
Innovation Solution
A neural network-type image processing device with fully connected units and intermediate layers, where connection coefficients are updated using a back propagation method, allowing for the analysis of image processing modules and their impact on the output image, enabling users to adjust the structure and optimize image processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple types of image processing are performed on input images to obtain desired output images, then the quality and suitability of output images for appearance inspection is improved, but the complexity of determining combinations and sequences of image processing increases significantly
Solution Approach 1:
The system performs self-learning through automatic learning processes where the learning unit automatically determines optimal combinations and sequences of image processing modules by learning from training data, eliminating the need for manual configuration by users
Solution Approach 2:
The system changes parameters automatically through learning by adjusting connection coefficients and processing parameters based on training data, transforming the static configuration process into a dynamic learning process that adapts to optimize image processing sequences
2Manufacturing precision
If trial and error method is used to determine image processing combinations, then desired output images can be obtained, but the time and burden required for configuration increases
Solution Approach 1:
The system performs preliminary learning actions by pre-learning optimal image processing combinations and sequences from training data before actual use, so that when processing inspection images, the system can directly apply pre-determined optimal configurations without time-consuming trial and error
Solution Approach 2:
The system implements feedback mechanisms where processing results are evaluated and fed back to the learning unit, which then adjusts connection coefficients and processing parameters to improve future performance, creating a continuous improvement loop that reduces configuration time
Data Source
AI summary
This neural network-type image processing device is provided with an input layer which comprises one unit where an input image is inputted, an output layer which comprises one unit where an output image is outputted, and multiple intermediate layers which are arranged between the input layer and the output layer and each of which comprises multiple units, the unit of the input layer, the units of the intermediate layers, and the unit of the output layer are fully connected with connection coefficients. The units of the intermediate layers are image processing modules which perform image processing on the image inputted to said units. The input image is inputted from the unit of the input layer, passes through the units of the intermediate layers, and is then outputted as an output image from the unit of the output layer; the connection coefficients are updated with learning based on a backpropagation algorithm.