Simulation of Image Weak Feature Self enhancement Based on Multimodal Neural Networks
With the rapid development of computer technology,the application of image recognition technology is increasingly widespread,and the demand for feature enhancement of target images is further increasing.In order to solve the problems of current image feature enhancement algorithms such as poor feature extraction ability and high image noise,this paper proposes a multimodal neural network enhancement algorithm based on attention mechanism.The algorithm first inputs image data,text description data,and similar dataset data as multimodal data,and uses con-volution and linear transformation to adjust them to the same dimension;Then,a neural network interaction module is used for feature fusion;The attention mechanism module is used to enhance information exchange between local adja-cent channels,and the pooling layer module is used to enhance target features;Finally,the image feature output se-quence is obtained by connecting the long and short term memory network to achieve the effect of feature enhance-ment.The experimental results show that the proposed algorithm improves the peak signal to noise ratio by 8.39%,improves the edge protection index by 5.15%,and improves the self enhancement ability of weak features.
NETL
NSTL
万方数据
