Image Annotation Optimization Method of Civil Engineering Supervision Based on Association Rule Mining Algorithm

The rapid development of civil engineering, leading to the vigorous development of civil engineering supervision industry, and civil engineering supervision, as an important mechanism to ensure that civil engineering to meet the relevant laws and regulations, its supervision data plays a vital role. Therefore, it is necessary to strengthen the importance of civil engineering supervision, and on this basis to improve the quality of the project. Because of its intuitive image and easy to interpret, efficient management and retrieval methods of civil engineering supervision image information have gradually become a research hotspot. This paper mainly discusses the problems existing in the image processing of civil engineering supervision, and puts forward the corresponding solutions, so as to provide some reference for the subsequent engineering application. At present, the requirements of high-level semantic expression of images are constantly improving, but the traditional solution is to manually annotate images, which is not only time-consuming, laborious and easy to cause ambiguity. Therefore, this paper studies the automatic annotation of civil engineering supervision images. In order to obtain the regional features of the image, the Mean Shift segmentation (MS) algorithm is proposed to segment the image. In order to improve the accuracy of automatic image annotation, the automatic annotation of civil engineering supervision image based on association rules is proposed, and the semantic extraction of civil engineering supervision image is carried out by Apriori algorithm. The experimental results show that the recall rate and precision rate of the proposed method are 85% and 91% respectively, and the precision rate is increased by 116% and the recall rate is increased by 117% compared with the CBA method. Compared with PLSA algorithm, the precision is increased by 15%, and the recall is increased by 13%. It also has a good effect on the image retrieval of civil engineering supervision.