| Title |
Research on Automatic Annotation Algorithm of Incomplete Image based on Mobile Computing Environment |
| DOI |
https://doi.org/10.5573/IEIESPC.2023.12.3.206 |
| Keywords |
Mobile computing; Incomplete image; Automatic labeling; Similarity measure |
| Abstract |
Automatic labeling of incomplete images can reduce the impact of image defects on understanding image content, which has great research value. The traditional image annotation method is manual annotation, which is inefficient, heavy, and subjective work. Therefore, an automatic image labeling algorithm based on mobile computing environment is proposed. The algorithm involves image preprocessing, feature extraction, similarity measurement of image features, and model training. In the preprocessing, image fragments are removed to reduce the complexity of the algorithm and increase the accuracy of automatic annotation. The image feature extraction was implemented by the scaling invariant feature transformation (SIFT) algorithm, and the image annotation algorithm was constructed based on the similarity between tag words and an image. Experimental results show that the recall rate of the algorithm reaches 97%, and the standard deviation of the normal distribution of automatic labeling after 300 iterations was 11. The results show that compared with the existing image automatic annotation methods, the proposed image automatic annotation algorithm has higher accuracy and better performance. |