Search

The method of detecting the whole tampered image increases many unnecessary calculations. In order to reduce the computational complexity and further improve the detection accuracy, this paper proposed a copy-move forgery detection method based on improved saliency map and local feature matching. Firstly, it combined the gradient of image to improve the saliency map, and separated the local salient regions containing high texture information of the image. Secondly, it only used SIFT（Scale Invariant Feature Transform） algorithm to extract feature points in this local area. Then, it adopted density clustering and two-stage matching strategy for images with low saliency, and adopted the strategy of superpixel segmentation and salient block feature matching for images with high saliency. Finally, it combined PSNR and morphological operations to locate the tampered area. Experiments on two public datasets show that the average detection time of this method is less than 10s, and the average detection accuracy is greater than 97%,which are better than the compared methods. The experimental results show that this method can greatly reduce the detection time, effectively improve the detection accuracy, and has good robustness to geometric transformation and post-processing operations. [ABSTRACT FROM AUTHOR]

Traditional manually designed feature descriptors are prone to cause a large number of mismatches in the face of multi temporal remote sensing images with large geometric and radiation differences and obvious changes in ground features. This paper developed a convolutional neural network matching algorithm with double screening and proposed double constraints to extract the deep semantic feature map of the image through CNN, it selected the key points that satisfied both the priority maximum principle and the precise extreme value principle from the feature map, and extracted 512-dimension descriptors from the corresponding positions. In the feature matching stage, the strategy of combining the rough to fine reverse matching constraints and RANSAC constraints ensured that the false matching points were eliminated effectively, and retained a certain number of correct matching points, which improved the matching accuracy. The experimental results show that the number of matches is moderate, the correct matching rate is improved compared with other algorithms selected in the experiment, and the image registration error is reduced by about 2 pixels. The running time is obviously improved, matching effect is good, with certain adaptability and robustness. [ABSTRACT FROM AUTHOR]

Complex event processing is a technology for analyzing the event streams in a dynamic environment. Complex event processing technology is usually implemented based on finite state automaton. During the matching process, a large number of overlapping partial matches will be generated by the event stream. The finite state automaton needs to maintain a large number of repeated matching states, which leads to the problem of redundant calculation in the methods based on this technology. In order to improve the matching efficiency of complex event processing, this paper proposes a method of using complex event instance coverage technology to realize complex event processing. By designing a temporary matching chain partition storage structure and matching algorithms based on this structure, redundant calculations can be reduced using complex event instance coverage, thereby achieving an improvement in matching efficiency. Experiments are performed on simulated and real datasets, and compared with two commonly used complex event processing technologies. The experimental results show that the proposed method can effectively reduce the redundant computation in the matching process while ensuring the correctness of the matching, and improve the overall matching efficiency. [ABSTRACT FROM AUTHOR]

To solve the problems of low efficiency and poor stability of image matching with angle and scale changes, this paper proposed an improved mesh motion statistical feature point screening algorithm based on local clustering. Firstly, as seed points by local region suppression algorithm could screen feature points with high response intensity and paired appearance, and with seed points as the clustering center to obtain the minimum enclosing rectangle as the moving grid could segment images. Then, it divided the motion grid into 3 x 3 neighborhood supporting estimator grids, and it could calculate the maximum gradient value of the motion grid in eight directions as the main direction of the motion grid. Finally, it rotated the neighborhood support estimator grid of the image to be matched to the main direction of the target image motion grid, and the mesh motion statistics algorithm usually screened the matches. The experimental results show that the matching accuracy rate of the proposed algorithm is above 90% for matching images with JPEG compression transform, light change and fuzzy transform. For images with rotation and scale transformation, the matching accuracy of the proposed algorithm is about 10% higher than that of the moving grid statistic algorithm, up to more than 70%. The algorithm takes only 13 min and has high efficiency. It shows that the proposed algorithm can screen the correct matching points stably and efficiently. [ABSTRACT FROM AUTHOR]

To improve the robustness of the existing multi-spectral image registration algorithms, this paper proposed a new algorithm based on multi-scale support region descriptors. Firstly, the algorithm extracted the Harris corner points as feature points. Secondly, it constructed the descriptor by combining the edge direction histograms calculated respectively in the support regions of different sizes around a feature point. Then, the similarity criterion was the Euclidean distance, and it obtained the initial matches by the ratio method. Finally, this paper proposed an outlier removal algorithm based on RANSAC algorithm. The experimental results show that the proposed algorithm can match multi-spectral images effectively, be more robust than the existing algorithms, and obtains more matches that are correct. [ABSTRACT FROM AUTHOR]

In order to fuse heterogeneous data with high precision, then express the buildings roof and facade information fully, this paper proposed a registration method of terrain laser scanning point cloud and image matching point cloud based on building footprint features. It extracted image matching point cloud building roof edge by boundary estimation, got terrain laser scanning point cloud building eaves boundary by a-shape algorithm, completed two footprint point cloud' s coarse registration by principal component analysis algorithm, centroid constraint and Rodrigues' s formula, finished fine registration according to the ICP algorithm. Experimental results show that the method can combine the advantages of cross-modal data, and improve the calculation efficiency and registration accuracy of image and point cloud registration effectively. [ABSTRACT FROM AUTHOR]

To solve the problems that the blur in the mosaic image and the ghosting appears after motion image mosaic. In the stage of image registration, the method used the perceptual hash similarity metric to select the weight ωi,j with the best transformation effect for calculating the local projection transformation matrix. Then, it used the frame difference method and the region growing algorithm to segment the moving object. The difference region is single sampled while the other regions were linear weighted fusion. Finally, this paper constructed a special aerial images dataset for image mosaic and conducted experiments based on it. The experimental results show that the proposed method requires low extra computation time and effectively suppress the blur and image abnormality in overlapping area, eliminate ghosting of images with moving objects, which significantly improve the quality of UAV’s aerial image mosaic. [ABSTRACT FROM AUTHOR]

The iterative closest point( ICP) algorithm is usually used in the fine registration of point cloud because of its high registration accuracy, but its registration accuracy and iteration convergence depend on the initial registration position of the point cloud to be registered. This paper proposed an algorithm of spatial optimization transformation matrix combining genetic algorithm with spatial distribution entropy, in which used a new point cloud spatial position evaluation method as the objective function of genetic algorithm, it used the genetic operator to guide the solution of search direction. The algorithm achieved the minimum spatial distribution entropy through the new population of iterations, after which the coarse registration of point cloud was achieved by decoding the optimal individuals. The experimental results show that the algorithm is effective and feasible, and it can overcome the problem that traditional method can’t provide a good initial position when the defects and noise exist in the point cloud. The algorithm can directly realize point cloud registration within the acceptable error range. [ABSTRACT FROM AUTHOR]