In this paper, a novel multi-sensor clustering algorithm, based on the density peaks clustering (DPC) algorithm, is proposed to address the multi-sensor data fusion (MSDF) problem. The MSDF problem is raised in the multi-sensor target detection (MSTD) context and corresponds to clustering observations of multiple sensors, without prior information on clutter. During the clustering process, the data points from the same sensor cannot be grouped into the same cluster, which is called the cannot link (CL) constraint; the size of each cluster should be within a certain range; and overlapping clusters (if any) must be divided into multiple clusters to satisfy the CL constraint. The simulation results confirm the validity and reliability of the proposed algorithm.

Sampling loss of the structural information of the image for the one-dimensional compression and bring about the loss of recognition accuracy, we propose the concept of two-dimensional compression samples. Using a set of sparse-based perception to get the sparse data on the raw data of the defect, fabric defect two-dimensional sparse. Finally, use of norm optimization method accurately decrypt the compressed data, the eigenvalues of different fabric defect classification. This approach solves the proliferation of data collection and the sensor waste greatly reduces the computational complexity, fabric defect classification, and thus to lay a theoretical foundation for machine vision to identify fabric defects.

Shadow detection and removal is a key problem in computer vison. Shadow can lead to loss or disruption of computer vision information, and then results in unstable situations or even failure of edge extraction, object recognition and image matching in image processing. The thesis concentrates on the research of moving object shadow detection and removal, and proposed the shadow detection and removal algorithm based on RGB and YCbCr color space. The experimental results show the algorithm based on the YCbCr color space have the lowly advantage.

Abstract. The present study on video detection is mianly based on image and video sequence. In the study of the video stream, goal shot play an important role, often indicates the highlight. In recent years, audiohas become more and more important with its rich information. Through analyzing the audio in order to find the excitement and bass, combining with the goal shot to determine whether the penalty shot or not. Experiments show that based on audio and video has higher recall and precision rate. 音频与球门融合的足球视频点球镜头检测

As the development of intelligent power system, much importance and requirement have been attached to the load data analysis. In view of the current rough classification of load data, propose a weighted fuzzy clustering algorithm to detail the load classification dividing, which adds a weight distribution process to balance the different influence of various factors. In addition, Two group of experiments are set to verify the efficiency of this method. The experiment results show that the algorithm is effective to accurately cluster the load data and supportive to the fine analysis of load data.