基于彩色图像分割的机器人足球目标识别

研究了机器人足球视觉系统中基于彩色图像分割的目标识别的方法。为了适应光照条件的变化，采用分离出亮度信息的YUV颜色空间；将彩色图像分割分为离线的颜色分类和实时的分割、识别两个部分，并采用最大似然法完成颜色的自动分类，满足了机器人足球视觉系统实时、准确的要求。试验证明，在光照条件改变的情况下能够有效地进行目标识别。     

Automatic recognition vision system guided for apple harvesting robot

In apple harvesting robot, the first key part is the machine vision system, which is used to recognize and locate the apples. In this paper, the procedure on how to develop an automatic recognition vision system guided for apple harvesting robot, is proposed. We first use a color charge coupled device camera to capture apple images, and then utilize an industrial computer to process images for recognising fruit. Meanwhile, the vector median filter is applied to remove the color images noise of apple, and images segmentation method based on region growing and color feature is investigated. After that the color feature and shape feature of image are extract, a new classification algorithm based on support vector machine for apple recognition is introduced to improve recognition accuracy and efficiency. Finally, these procedures proposed have been tested on apple harvesting robot under natural conditions in September 2009, and showed a recognition success rate of approximately 89% and average recognition time of 352 ms.     

A new pattern representation scheme using data compression

We propose the PRDC (Pattern Representation based on Data Compression) scheme for media data analysis. PRDC is composed of two parts: an encoder that translates input data into text and a set of text compressors to generate a compression-ratio vector (CV). The CV is used as a feature of the input data. By preparing a set of media-specific encoders, PRDC becomes widely applicable. Analysis tasks - both categorization (class formation) and recognition (classification) - can be realized using CVs. After a mathematical discussion on the realizability of PRDC, the wide applicability of this scheme is demonstrated through the automatic categorization and/or recognition of music, voices, genomes, handwritten sketches and color images     

Two stage packet classification using most specific filter matching and transport level sharing

In this paper we introduce two new concepts to the design of packet classification systems. First, we propose most specific filter matching (MSFM), an improvement over the well known Cross Producting algorithm [V. Srinivasan, S. Suri, G. Varghese, M. Waldvogel, Fast and scalable layer four switching, in: Proceedings of ACM SIGCOMM, 1998] that significantly reduces the memory requirement of the earlier scheme. Second, we suggest that rules specifying the same source–destination IP prefix pair can be grouped together forming shared sets of transport level fields. This property of Transport Level Sharing (TLS), which characterizes real world classification databases is exploited for reducing a classifier’s memory requirement and for hardware acceleration.We split the classification process into two stages. First, we perform classification on source–destination IP prefix pairs using the MSFM algorithm. Second, we perform classification on transport level fields exploiting transport level sharing. It is the combination of most specific filter matching and transport level sharing which results in a scheme that requires no more than 11 dependent memory accesses in the critical path independent of the size of the classification database. The memory access bandwidth of our scheme is also bounded when our scheme is accelerated in hardware. Compared to other schemes which involve a small and predictable number of steps in the critical path (e.g., Cross Producting [V. Srinivasan, S. Suri, G. Varghese, M. Waldvogel, Fast and scalable layer four switching, in: Proceedings of ACM SIGCOMM, 1998] or Recursive Flow Classification [P. Gupta, N. McKeown, Packet classification on multiple fields, in: Proceedings of ACM SIGCOMM, 1999]) the combination of most specific filter matching and transport level sharing is associated with the least memory requirement.     

MULTICLASS OBJECT RECOGNITION USING CLASS-CONDITIONAL INDEPENDENT COMPONENT ANALYSIS

A new modeling technique, based on independent component analysis (ICA), is proposed to represent and recognize high-dimensional samples from a large set of classes. The model is constructed via density estimation techniques, and recognition is performed in the Bayesian decision framework. We show that the technique can be successfully used for automatic object identification in environments where a visual observer is faced with a classification problem in high-dimensional spaces with a large number of classes. A first experiment illustrates that classification using an ICA representation is a technique that, even in low dimensions, performs comparably to standard classification techniques. The second experiment tests the ICA classification model on high-dimensional data. Recognition was performed using local color histograms of images corresponding to 400 different objects. It is also shown how our approach outperforms other techniques commonly used in the context of appearance-based recognition.     

Multi-stage classification of emotional speech motivated by a dimensional emotion model

This paper deals with speech emotion analysis within the context of increasing awareness of the wide application potential of affective computing. Unlike most works in the literature which mainly rely on classical frequency and energy based features along with a single global classifier for emotion recognition, we propose in this paper some new harmonic and Zipf based features for better speech emotion characterization in the valence dimension and a multi-stage classification scheme driven by a dimensional emotion model for better emotional class discrimination. Experimented on the Berlin dataset with 68 features and six emotion states, our approach shows its effectiveness, displaying a 68.60% classification rate and reaching a 71.52% classification rate when a gender classification is first applied. Using the DES dataset with five emotion states, our approach achieves an 81% recognition rate when the best performance in the literature to our knowledge is 76.15% on the same dataset.