Aili Wang

Object Tracking Based on Hierarchical Convolutional Features

A novel object tracking algorithm based on hierarchical convolutional features was proposed in this paper. Firstly, the tracking algorithm uses the hierarchical networks of VGG-Net-19 to extract the hierarchical convolutional features of image, having a greater improvement than using only one layer to do that. Secondly, the algorithm obtains features by using correlation filtering method with weighted fusion, so as to determine the real position of the target according to the characteristics of different layers. The experimental results show that, compared with the current four popular object tracking algorithms, the proposed algorithm achieves better accuracy and success rate, and the results are consistent in OPE (one-pass evaluation), SRE (spatial robustness evaluation) and TRE (temporal robustness evaluation).

Automatic Polyp Detection from Endoscope Image using Likelihood Map based on Edge Information.

An endoscope is a medical instrument that acquires images in side the human body. This paper proposes a new approach for the automatic detection of polyp regions in an endoscope image by generating a likelihood map with both of edge and color information to obtain high acc ura y so that probability becomes high at around polyp candidate region. Next, Histograms of Oriente d Gradients (HOG) features are extracted from the detected region and random forests are applied for the cl assification to judge whether the detected region is polyp region or not. It is shown that the proposed approach has high accuracy for the polyp detection and the usefulness is confirmed through the computer experiment s with endoscope images.

Shape Recovery of Polyp from Endoscope Image Using Blood Vessel Information

Endoscope is used to remove the polyp in the medical diagnosis. Absolute size of polyp has been usually estimated by medical doctor with their empirical judgement using endoscope. However this estimation depends on the experience and skill of medical doctor and it is sometimes necessary to use the medical thread with known size for estimating the size of polyp. This paper aims to help medical doctor by proposing a new approach to estimate the size and 3D shape of polyp as a medical supporting system. This proposed approach uses blood vessel as a target with a known size to estimate the absolute size of polyp. Using sequential two images make it possible to estimate the movement of endoscope and reflectance parameter. The idea of using blood vessel is the key idea of this paper, where color information, labeling, morphology processing are used estimate the size and 3D shape of polyp as a final goal. Experiments with endoscope images are demonstrated to evaluate the validity of proposed approach.

3D Shape from SEM Image Using Improved Fast Marching Method

This paper proposes an improved fast marching method to recover 3D shape from a Scanning Electron Microscope (SEM) image as a Shape from Shading approach. First, the method uses the second-order finite difference and the information of diagonal grid points to obtain highly accurate solution. Then the method speeds up with increasing the number of the neighboring points, and changes the update mode to avoid sorting processing. Finally, the results were compared between proposed method and previous method via simulation and real SEM image. Experimental results show the proposed method gives the better and faster 3D shape.

Teaching Reform and Innovation of Communication Principles Curriculum Based on O2O Mode

Since the communication principles content are abstract and complex, and systematic, students generally reflect this course is difficult to learn, understand and master, teachers also feel difficult to teach. Therefore, how to improve the quality of teaching of communication principles curriculum is the key problem in teaching process. The Online to Offline (O2O) hybrid teaching mode can realize classroom flip, further deepen the curriculum reform, make curriculum quality standards, recard and optimize teaching content, refine teaching requirements, improve teaching methods and the construction of curriculum resources, curriculum construction quality assurance system. It explores the reform of other courses for Communication Engineering Department, provides technical support and accumulates valuable experience for subsequent courses to complete O2O mode teaching.

A novel human detection algorithm combining HOG with LBP histogram Fourier

Human detection is a key problem in computer vision, with several applications that have the potential to positively impact quality of life. In recent years, the number of approaches to detect human in video sequences has grown steadily. This paper proposed a novel approach to improve the capability of human detection. After investigating the image texture analysis methods, we adopt the new good texture descriptor, Local Binary Pattern histogram Fourier (LBPHF). Then we combine the LBPHF with the Histograms of Oriented Gradients (HOG) as feature sets, and use linear SVM to train our classifier. In the experiment on the INRIA personal dataset which is well known relatively good human detections dataset, it is shown that our detector combining with the LBPHF significantly outperforms the other methods. Moreover, the time cost is much less and the dimension is reduced.