This research suggests a deep learning-based method for text identification
from hazy images using a self-collected dataset. The problem of identifying text from
hazy images is challenging due to the degradation of the image quality caused by
various atmospheric conditions. To address this issue, the proposed approach utilizes a
deep learning framework that comprises a hybrid architecture wherein a convolutional
neural network (CNN) is employed for feature extraction and a recurrent neural
network (RNN) is utilized for sequence modelling. A self-collected dataset is employed
for training and validation of the proposed approach, which contains hazy images of
various text sizes and fonts. The experimental findings show that the suggested
technique outperforms state-of-the-art approaches in correctly recognizing text from
hazy images. Additionally, the proposed self-collected dataset is publicly available,
providing a valuable resource for future investigations in the field. Overall, the
proposed approach has potential applications in various domains, including image
restoration, text recognition, and intelligent transportation systems. The performance of
the trained model is then evaluated using a third-party dataset consisting of blurry
photos. The effectiveness of the model may be evaluated using standard metrics,
including accuracy, precision, recall, and F1-score.
Keywords: Convolutional neural network, Hazy images, Image dehazing, Selfcollected dataset, Text identification.
