高慧,曾庆尚,韩明峰.基于梯度直方图与密度度量模型的图像伪造检测算法[J].包装工程,2017,38(23):205-210.
GAO Hui,ZENG Qing-shang,HAN Ming-feng.Image Forgery Detection Algorithm Based on Gradient Histogram and Density Measurement Model[J].Packaging Engineering,2017,38(23):205-210. |
| 基于梯度直方图与密度度量模型的图像伪造检测算法 |
| Image Forgery Detection Algorithm Based on Gradient Histogram and Density Measurement Model |
| 投稿时间：2017-06-14 修订日期：2017-12-10 |
| DOI： |
| 中文关键词: 图像伪造检测 彩色图像 Hessian算法 四级窗口 梯度直方图 密度度量模型 |
| 英文关键词: image forgery detection color image Hessian algorithm four-level window gradient histogram density measurement model |
| 基金项目:国家自然科学基金（61170224）；山东省自然科学基金（ZR2012FL07） |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 为了解决当前图像伪造检测算法在内容识别过程中易丢失色彩信息而导致不理想的检测精度与鲁棒性等问题，提出基于梯度直方图耦合密度度量模型的图像伪造检测算法。方法 首先引入RGB彩色图像映射模型，求取图像的颜色不变量。将图像的颜色不变量作为输入量，利用 算法检测图像的特征点。然后以特征点为中心构造四级窗口，通过求取窗口内梯度累加值，形成低维度的特征描述符，并利用特征点对应的梯度直方图构造相似性度量模型进行特征点匹配。最后借助欧式距离，构造密度度量模型，对特征点进行归类，以完成伪造检测。结果 仿真实验表明，与当前图像伪造检测算法相比，所提算法具有更高的检测正确度，高达99.6%。结论 所提算法具有较高的伪造检测精度与鲁棒性，在图像信息、包装印刷等领域具有良好的应用价值。 |
| 英文摘要: |
| The work aims to put forward an image forgery detection algorithm based on gradient histogram coupling density measurement model, for the purpose of solving the problem of poor detection accuracy and robustness induced by the current image forgery detection algorithm that easily loses the color information in the process of identifying the contents. Firstly, the RGB color image mapping model was introduced to obtain the color invariants of the image. With the image color invariants as the inputs, the Hessian algorithm was used to detect the feature points of the image. Then, the four-level window was constructed with feature points as the center, and the low dimensional feature descriptor was formed by finding the gradient accumulation value in the window; and the similarity measurement model was constructed with the gradient histogram corresponding to the feature points to match the feature points. Finally, the Euclidean distance was used to construct the density measurement model, which was used to classify the feature points, and then the forgery detection was completed. The simulation results showed that, compared with the current image forgery detection algorithm, the proposed algorithm had higher detection accuracy which was up to 99.6%. The proposed algorithm has higher forgery detection accuracy and robustness, and it has good application value in image information, packaging, printing and other fields. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
渝公网安备 50010702501716号