| 黄 冠,江晓昱,胡章威,周 萍,曾敬贤.模糊与噪声对人耳识别性能的影响[J].,2018,(5):863-866 |
| 模糊与噪声对人耳识别性能的影响 |
| Impact of Blur and Noise on the Ear Recognition Performance |
| 投稿时间：2017-06-29 修订日期：2017-07-23 |
| DOI：10.13241/j.cnki.pmb.2018.05.013 |
| 中文关键词: 耳图像识别 模糊 热噪声 |
| 英文关键词: Ear recognition Blur Noise |
| 基金项目: |
|
| 摘要点击次数: 425 |
| 全文下载次数: 244 |
| 中文摘要: |
| 摘要 目的：评价模糊与噪声对不同图像识别算法在人耳图像识别过程中的影响。方法：对500 个清晰图像进行模糊和热噪声处理，比较方向梯度直方图(HOG)、局部相位量化(LPQ)和局部二值模式(LBP)三种不同的特征提取算法对生物识别性能的影响。结果：HOG、LPQ和LBP算法对清晰人耳图像的识别率都很高，在无噪声和模糊等信号衰减的情况下，识别率分别达到了85.96 %、95.62 %和91.36 %。在这三种算法中，模糊和热噪声对人耳识别能力均有不同程度的下降。热噪声对人耳识别能力的影响显著高于模糊，但是当模糊和热噪声同时存在图像中时，模糊和噪声会互相加强，导致三种人耳识别算法均不能获得好的识别结果。结论：LPQ算法对模糊有较好的识别性，并且在人耳图像的获取和处理过程中可尽量减少热噪声的产生。 |
| 英文摘要: |
| ABSTRACT Objective: To evaluate the impact of blur and noise on ear recognition performance. Methods: 500 ear images were collected and treated with Gaussian blur and noise in this work. Three feature extraction algorithms, Histogram of Gradient (HOG), Local Phase Quantization (LPQ) and Local Binary Pattern (LBP) were compared for the ear recognition performance. Results: The recognition rate of HOG, LPQ, and LBP were relatively high with 85.96 %, 95.62 % and 91.36 % respectively without blur and noise. The influence of thermal noise on the recognition ability of ear was significantly higher than that of the blur. However, the noise and blur will strength each other when they exist in ear images at the same time. Conclusion: LPQ algorithm was robust to blur. It is proper to decrease the noise in order to obtain better ear recognition results. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
