Current Issue : April - June Volume : 2019 Issue Number : 2 Articles : 5 Articles
RGB-D cameras offer both color and depth images of the surrounding environment, making\nthem an attractive option for robotic and vision applications. This work introduces the BRISK_D\nalgorithm, which efficiently combines Features from Accelerated Segment Test (FAST) and Binary\nRobust Invariant Scalable Keypoints (BRISK) methods. In the BRISK_D algorithm, the keypoints are\ndetected by the FAST algorithm and the location of the keypoint is refined in the scale and the space.\nThe scale factor of the keypoint is directly computed with the depth information of the image. In\nthe experiment, we have made a detailed comparative analysis of the three algorithms SURF, BRISK\nand BRISK_D from the aspects of scaling, rotation, perspective and blur. The BRISK_D algorithm\ncombines depth information and has good algorithm performance....
In order to resolve the problem of the image degradation, an image enhancement\nmethod based on fractional calculus and Retinex is proposed,\nwhich can preserve or enhance texture information and remove the noise of\nimages. The fractional differential is used to preprocess the input image to\nenhance texture information, and using guided filter to estimate the illumination\ncomponent, so it has less halo phenomena. The reflection component,\nobtained according to the Retinex theory, is denoised by fractional integral to\nremove the noises. The image is equalized by the contrast limited adaptive\nhistogram equalization to improve the image contrast, and a final enhanced\nimage is obtained. The experimental results show that the method can effectively\nachieve image enhancement, and the enhanced image has better visual\neffects....
Scene classification of high-resolution remote sensing (HRRS) image is an\nimportant research topic and has been applied broadly in many fields. Deep\nlearning method has shown its high potential to in this domain, owing to its\npowerful learning ability of characterizing complex patterns. However the\ndeep learning methods omit some global and local information of the HRRS\nimage. To this end, in this article we show efforts to adopt explicit global and\nlocal information to provide complementary information to deep models.\nSpecifically, we use a patch based MS-CLBP method to acquire global and local\nrepresentations, and then we consider a pretrained CNN model as a feature\nextractor and extract deep hierarchical features from full-connection\nlayers. After fisher vector (FV) encoding, we obtain the holistic visual representation\nof the scene image. We view the scene classification as a reconstruction\nprocedure and train several class-specific stack denoising autoencoders\n(SDAEs) of corresponding class, i.e. , one SDAE per class, and classify the test\nimage according to the reconstruction error. Experimental results show that\nour combination method outperforms the state-of-the-art deep learning classification\nmethods without employing fine-tuning....
In most visual tracking tasks, the target is tracked by a bounding box given in the first\nframe. The complexity and redundancy of background information in the bounding box inevitably\nexist and affect tracking performance. To alleviate the influence of background, we propose a robust\nobject descriptor for visual tracking in this paper. First, we decompose the bounding box into\nnon-overlapping patches and extract the color and gradient histograms features for each patch.\nSecond, we adopt the minimum barrier distance (MBD) to calculate patch weights. Specifically,\nwe consider the boundary patches as the background seeds and calculate the MBD from each patch\nto the seed set as the weight of each patch since the weight calculated by MBD can represent the\ndifference between each patch and the background more effectively. Finally, we impose the weight\non the extracted feature to get the descriptor of each patch and then incorporate our MBD-based\ndescriptor into the structured support vector machine algorithm for tracking. Experiments on two\nbenchmark datasets demonstrate the effectiveness of the proposed approach....
In order to reduce the computational complexity of searching in massive information\nin detecting of warhead targets, background removal is usually the\nfirst step of target detection algorithm in sequential frame images. In this paper,\nan adaptive multi-exposure time preserving star edge small area filtering\nbackground removal algorithm is proposed, which can suppress the background\nand preserve the target and star edges. This algorithm not only ensures\nthe accuracy of centroid and orbit determination, but also reduces false\nalarm and improves tracking accuracy....
Loading....