Search

This paper presents a new scheme for off-line recognition of totally unconstrained handwritten numerals. The work is motivated by two approaches: Affinity Propagation (AP) clustering and Biomimetic Pattern Recognition (BPR). It can relatively classify the samples by calculating the distance to the relative subspace. The training sample space is constructed by the AP algorithm and BPR theory. The posterior probabilities based on the class condition are estimated to reduce the rejection caused by the space overlapping with low misclassification. Experiments have been performed with Concordia University CENPARMI's handwritten digit database and Nanjing University of Science and Technology's handwritten amount database. Experimental results indicate that the proposed classifier has a higher recognition rate than the traditional classifiers.

An improved quantum genetic algorithm (IQGA) is proposed for path planning of mobile robot in unknown environment, which uses problem-specific quantum genetic algorithm for robot path planning instead of the standard genetic algorithm(GA) and quantum genetic algorithm(QGA). Six genetic operators operated chromosomes and improved fitness function to evaluate new individuals, and this algorithm is capable of finding an optimal or near-optimal robot path in both complex static and dynamic environments. The simulation results show that the proposed method is accurate and effective.

Time-lapse microscopy imaging serves as an important method to measure the cycle progression of individual cells in a large population. The lacking of automatic systems that can deal with large volume of time-lapse optical microscopy imaging is a bottleneck of modern bio-imaging research. In this paper we present an effective computerized method for overlapping cell nuclei detection in time-lapse image sequences. We design a two-stage detection scheme for such cells in each image of the image sequence. Features of the skeleton and the contour of the cell model are extracted in the two stages, and overlapped cells are detected by the similarity of the features with the training samples. Experimental results shows the proposed method is very effective and the work shows great assistance for the cell nuclei segmentation afterward.

In this paper, a new particle swarm based stereo algorithm is presented. Our motivation is to improve the accuracy of the disparity map by removing the mismatches caused by both occlusions and false targets. In our approach, the stereo matching problem is divided into two steps, including partial matching of segmented image and particle swarm optimization of the rest. The algorithm first takes advantage of SAD and Dynamic Programming to remove the mismatches mainly caused by visibility problems; after the first step, the algorithm selects all the rest image segmented regions, takes them as a particle and uses particle swarm to optimization it. In the second step, the cost function is defined on the pixel level, as well as on the segmented level, while the pixel level measures the data similarity based the current disparity map, the segmented level incorporates a smooth term. Results obtained for benchmark indicate that the proposed method is able to get rather accurate disparity maps.

By the idea of manifolds learning, this paper presents a new method of dimensionality reduction of high dimensional data. The trait of the method is to exploit image matrixes to directly construct the local scatter matrix and the nonlocal scatter matrix. Its discriminant criterion function is characterized by maximizing the difference between the nonlocal scatter and the local scatter after the samples are projected. The new method is called the two-dimensional marginal discriminant projection (2DMDP). The new discriminant criterion is similar to the maximum margin criterion in form. The criterion main purpose is to find a projection direction (i.e. projection axes) that simultaneously maximizes the nonlocal scatter of projected sample, and minimizes the local scatter of projected sample. The experimental results on YALE face database and ORL face database show that the proposed method outperforms LPP and UDP in terms of recognition rate, and even outperforms LDA when the training sample size per class is small.

This paper presents a new method of dimensionality reduction of high dimensional data. The new discriminant criterion function be characterized by between the nonlocal scatter and the local scatter, and to directly construct between local scatter matrix and nonlocal scatter matrix by sample image matrixes. The criterion main purpose is to find a group of projection axis that simultaneously maximizes the nonlocal scatter and minimizes the local scatter of sample feature. The experimental results on YALE face database and AR face database show that the proposed method consistently outperforms LPP and UDP based on image vector, and even outperforms LDA.

Artificial fish swarm algorithm (AFSA) is a global optimization method proposed recently. After analyzing the disadvantages of AFSA, this paper introduced best-step operator and refined the prey behavior. An improved artificial fish-swarm algorithm for the RBF neural network and a model based on this method is developed. Finally the new algorithm is applied to the problem of expression recognition. The research indicates that the new algorithm has some advantages in terms of convergence performance, recognition rate and so on.

This research focuses on the problem of autonomous land vehicle (ALV) navigating on the unstructured roads which have no lane marks, may have degraded surfaces and edges, and may have shrub and weed alongside. Because of the complexity and irregularity of the road edges, the accurate and stable edges can not be obtained on the pixel level. We developed a novel sensor fusion and motion fusion parallel strategy on the feature level to model the unstructured road edges accurately. The road edges from grey images and lane cues from laser range finders (LRF) are integrated to form a consistent estimation of the edge position based on covariance intersection (CI) filter. The fused edges of consecutive frames are associated and integrated also based on CI. The quantitative evaluation method is also proposed to assess the quality of road edges. These algorithms are tested on a variety of complex unstructured roads, the autonomous navigation is successful and the average speed achieves 20 km/h. The spot navigation experiments show the effectiveness and efficiency of these algorithms to the accurate unstructured road tracking

A fast method for vanishing point (VP) estimation and tracking in road images and its application for lane detection are investigated in this paper. After determine the region of interest of the road images, we use `Sobel' operator to detect boundaries. Then the boundary image is transformed to the Hough Space by Hough Transform. We proposed a Gaussian Predication Model on the basis of the latest vanishing point to predicate the current VP, and we constructed an objective function including the linear segment intensity and the predicated VP, subsequently the optimal position of the VP is estimated through the least square method. The computational complexity of the estimation algorithm is O(n), which can operate in real time. We test our algorithm on many structured road and semi-structured road images, the results show that it can robustly determine the VP position regardless the disturbance of around pedestrian, crowd and vehicles. Moreover, experiments on PETS2001 dataset show its application to the road lane detection for the driver assistant system.

We present a study on the statistically uncorrelated optimal discriminant plane in this paper. A new concise formula for finding the statistically uncorrelated optimal discriminant plane has been developed, which overcomes the shortcoming of Jin's statistically uncorrelated optimal discriminant plane. The numerical experiments conducted on IRIS data verify the correctness of the above statements

In Chinese, thousands of bankchecks are validated due to manual processing every day to find out the bankchecks, which are invalidate to the bank's rules or contain error input information. It is necessary to relieve people from these intensive and tedious works. This paper presents a technique, which is the preprocess stage in our auto-validating system, for extracting the user-entered courtesy amount item from Chinese bankcheck images. Seal imprints in different types and positions cause the main difficulty in the extracting procedure. Instead of removing lines in bi-level images, we proposed a line removal algorithm based on crossing shape analysis in gray-level image. The handwritten information is extracted by a new recursive thresholding algorithm, which can continue removing the brighter background until only the darkest object is left. The proposed methods have been evaluated on real-life Chinese checks. And the results demonstrate the effectiveness of the proposed preprocessing techniques.

After the pavement is built up, to prolong its service life and reduce its maintenance expenses are the important things to consider. In order to make a rational maintenance plan, a thorough and accurate understanding in pavement surface situation is needed. In the past years, the measuring of the pavement surface state still depends on the handcraft or semiautomatic devices in China. These kinds of methods are not only extremely low in efficiency but also needs to close the transport. Nanjing University of Science and Technology (NJUST) developed an intelligent equipment for pavement surface measuring. Its main technical performance has already reached the level of the international similar products.

We present a new method for obstacle and pedestrian detection with lower computation complexity and higher detection performance on a small target. By calculating stereo disparity according to the road parameters, our system efficiently detects objects above the ground. Experiments on over 10,000 images captured in urban areas demonstrate our method's effectiveness. We also propose a multi-scale compatible pedestrian detector. Detection speed is improved by avoiding size adjustment of input. Experiments on three public pedestrian databases and our HENU database show that our detector achieves better results than the state-of-the-art detection quality at equal speed, especially on pedestrian target with small size.