Application Analysis of Image Enhancement Method in Deep Learning Image Recognition Scene

Abstract

Application analysis of the image enhancement method in deep learning image recognition scene is conducted in this paper. Generally speaking, scene recognition of natural scenes is relatively difficult due to the more complex and diverse environment. It is usually done through two steps: text detection and text recognition. To enhance the traditional methods, this paper integrates the deep learning models to construct the core efficient framework for dealing with the complex data. The text method uses a sequence recognition network based on a two-way decoder based on adjacent attention weights to recognize text images and predict the output. For the further analysis, the core systematic modelling is demonstrated. The proposed model is tested on the public data sets as a reference. The experimental verification has shown the result that the proposed model is efficient.