CNN Encoder and ConvLSTM Network for Image Difference Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying differences across a plurality of images face challenges in achieving high spatiotemporal resolution while minimizing memory and processing requirements, leading to an undue burden on computing resources.
Innovation Solution
The use of a system comprising a convolutional neural network (CNN) encoder connected with a convolutional Long Short-Term Memory (ConvLSTM) network to identify changes across images, generating feature maps and change maps that reduce memory requirements while maintaining high resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high spatial and temporal resolution are achieved by analyzing all image data, then identification precision is improved, but memory requirement and processing time increase significantly
Solution Approach 1:
The patent extracts only the essential information needed for change detection by processing images through a CNN encoder that generates feature maps, then feeding these compressed features into a ConvLSTM network. This extraction approach retains high spatiotemporal resolution while dramatically reducing the quantity of data that must be stored and processed, directly resolving the memory requirement contradiction.
Solution Approach 2:
The patent segments the image analysis process into distinct stages: first processing individual images through the CNN encoder to create feature representations, then analyzing temporal changes through the ConvLSTM network. This segmentation allows each component to operate on reduced-dimensional data, maintaining resolution while lowering overall memory and processing demands.
2Measurement precision
If high spatial and temporal resolution are achieved by analyzing all image data, then identification precision is improved, but processing time increases significantly
Solution Approach 1:
By extracting and processing only the essential feature information through the CNN encoder before temporal analysis, the system reduces the processing volume significantly. The ConvLSTM network then operates on compressed feature maps rather than raw image data, maintaining high temporal resolution while reducing processing time requirements.
Solution Approach 2:
The CNN encoder performs preliminary processing on each image to create feature maps that capture essential spatial information. This preliminary action prepares the data in advance for the temporal analysis stage, allowing the ConvLSTM network to process changes more efficiently without requiring re-processing of the entire image data, thus reducing overall processing time while maintaining resolution.
Data Source
AI summary
A computer-implemented method for identifying one or more changes across a plurality of images, the method comprising: receiving, at a convolutional neural network, CNN, encoder, CNN input data comprising data associated with each pixel of each of the plurality of images; propagating the CNN input data through the CNN encoder to generate a plurality of feature maps, wherein each feature map comprises a feature classification of each pixel of a respective image of the plurality of images according to a feature classification scheme, wherein the feature classification scheme comprises a plurality of classifications and is generated by the CNN encoder based on training data; receiving, at a ConvLSTM network, ConvLSTM input data comprising the plurality of feature maps generated by the CNN encoder; and propagating the ConvLSTM input data through the ConvLSTM network to generate a change map, wherein the change map comprises change data indicative of one or more changes across the plurality of images.