In order to achieve higher accuracy digital image information recognition,the study completes the recognition of image targets based on contour extraction and contour center-of-mass height incremental feature descriptor construction,and optimizes the descriptor similarity calculation using shape complexity analysis.In the validation experiments of the effectiveness of the digital media image target recognition method based on contour center-of-mass height incremental features,the retrieval performance of the pro-posed method of the study is better than several common methods proposed in the table,and the retrieval rate is improved by 5.69％～24.81％.The experimental results illustrate that the center-of-mass height increment descriptor accurately describes the position relationship between contour points and points,and provides better performance for distinguishing similar contours,and also verifies that the complexity evaluation of contours can help improve the confidence of matching results.In the image recognition experiments with noise,when the noise level increases above 0.6,although the retrieval accuracy shows a significant decrease,it can still main-tain a high recognition accuracy,which indicates that the method proposed in the study has a strong robustness to the interference of noise.
关键词
数字媒体/目标识别/轮廓特征/质心高度增量/边缘检测
Key words
digital media/target recognition/contour feature/center-of-mass height increment/edge detection
维普
万方数据