Jun-Wei Hsieh

Morphology-based license plate detection from complex scenes

This paper presents a morphology-based method for detecting license plates from cluttered images. The proposed system consists of three major components. At the first, a morphology-based method is proposed to extract important contrast features as guides to search the desired license plates. The contrast feature is robust to lighting changes and invariant to several transformations like scaling, translation, and skewing. Then, a recovery algorithm is applied for reconstructing a license plate if the plate is fragmented into several parts. The last step is to do license plate verification. The morphology based method can significantly reduce the number of candidates extracted from the cluttered images and thus speeds up the subsequent plate recognition. Under the experimental database, 128 examples got from 130 images were successfully detected. The average accuracy of license plate detection is 98%. Experimental results show that the proposed method improves the state-of-the-art work in terms of effectiveness and robustness of license plate detection.

Image Registration Using a New Edge-Based Approach

A new edge-based approach for efficient image registration is proposed. The proposed approach applies wavelet transform to extract a number of feature points as the basis for registration. Each selected feature point is an edge point whose edge response is the maximum within a neighborhood. By using a line-fitting model, all the edge directions of the feature points are estimated from the edge outputs of a transformed image. In order to estimate the orientation difference between two partially overlapping images, a so-called “angle histogram” is calculated. From the angle histogram, the rotation angle which can be used to compensate for the difference between two target images can be decided by seeking the angle that corresponds to the maximum peak in the histogram. Based on the rotation angle, an initial matching can be performed. During the real matching process, we check each candidate pair in advance to see if it can possibly become a correct matching pair. Due to this checking, many unnecessary calculations involving cross-correlations can be screened in advance. Therefore, the search time for obtaining correct matching pairs is reduced significantly. Finally, based on the set of correctly matched feature point pairs, the transformation between two partially overlapping images can be decided. The proposed method can tolerate roughly about 10% scaling variation and does not restrict the position and orientation of images. Further, since all the selected feature points are edge points, the restriction can significantly reduce the search space and, meanwhile, speed up the matching process. Compared with conventional algorithms, the proposed scheme is a great improvement in efficiency as well as reliability for the image registration problem.

Shadow elimination for effective moving object detection by Gaussian shadow modeling

Abstract This paper presents a novel approach for eliminating unexpected shadows from multiple pedestrians from a static and textured background using Gaussian shadow modeling. First, a set of moving regions are segmented from the static background using a background subtraction technique. The extracted moving region may contain multiple shadows from various pedestrians. In order to remove these unwanted shadows completely, a histogram-based method is proposed for isolating each pedestrian from the extracted moving region. Based on the results, a coarse-to-fine shadow modeling process is then applied for eliminating the unwanted shadow from the detected pedestrian. At the coarse stage, a moment-based method is first used for obtaining the rough shadow boundaries. Then, the rough approximation of the shadow region can be further refined through Gaussian shadow modeling. The chosen shadow model is parameterized with several features including the orientation, mean intensity, and center position of a shadow region. With these features, the chosen model can precisely model different shadows at different conditions and provide good capabilities for completely eliminating the unexpected shadows from the scene background. Due to the simplicity of the proposed method, all the shadows can be eliminated immediately (in less than 0.5 s). Experiments demonstrate that approximately 94% of shadows can be successfully eliminated from the scene background.

Video-Based Human Movement Analysis and Its Application to Surveillance Systems

This paper presents a novel posture classification system that analyzes human movements directly from video sequences. In the system, each sequence of movements is converted into a posture sequence. To better characterize a posture in a sequence, we triangulate it into triangular meshes, from which we extract two features: the skeleton feature and the centroid context feature. The first feature is used as a coarse representation of the subject, while the second is used to derive a finer description. We adopt a depth-first search (dfs) scheme to extract the skeletal features of a posture from the triangulation result. The proposed skeleton feature extraction scheme is more robust and efficient than conventional silhouette-based approaches. The skeletal features extracted in the first stage are used to extract the centroid context feature, which is a finer representation that can characterize the shape of a whole body or body parts. The two descriptors working together make human movement analysis a very efficient and accurate process because they generate a set of key postures from a movement sequence. The ordered key posture sequence is represented by a symbol string. Matching two arbitrary action sequences then becomes a symbol string matching problem. Our experiment results demonstrate that the proposed method is a robust, accurate, and powerful tool for human movement analysis.

Motion-based video retrieval by trajectory matching

This paper proposes a hybrid motion-based video retrieval system to retrieve desired videos from video databases through trajectory matching. The hybrid method includes a sketch-based scheme and a string-based one to analyze and index a trajectory with more syntactic meanings. First of all, this method uses a sampling technique to extract a set of control points from each trajectory as features. Then, the sketch-based method uses a curve fitting technique to interpolate some missed data in this set of control points. Then, the visual distance between any two trajectories can be directly measured by comparing their position data. The visual distance is good in solving the problem of translation-invariant trajectory matching but poor in solving the problem of partial trajectory matching. Therefore, in addition to the visual distance, the hybrid method uses the string-based scheme to compare any two trajectories according to their syntactic meanings. With the help of the syntactic distance, many impossible candidates can be filtered out in advance and thus the accuracy of video retrieval can be much enhanced. In addition, the problem of partial trajectory matching will become easy to be solved. Thus, even though a partial trajectory is queried, all desired video clips still can be very accurately retrieved. Experimental results have proved the superiority of our proposed method.

Symmetrical SURF and Its Applications to Vehicle Detection and Vehicle Make and Model Recognition

Speeded-Up Robust Features (SURF) is a robust and useful feature detector for various vision-based applications but it is unable to detect symmetrical objects. This paper proposes a new symmetrical SURF descriptor to enrich the power of SURF to detect all possible symmetrical matching pairs through a mirroring transformation. A vehicle make and model recognition (MMR) application is then adopted to prove the practicability and feasibility of the method. To detect vehicles from the road, the proposed symmetrical descriptor is first applied to determine the region of interest of each vehicle from the road without using any motion features. This scheme provides two advantages: there is no need for background subtraction and it is extremely efficient for real-time applications. Two MMR challenges, namely multiplicity and ambiguity problems, are then addressed. The multiplicity problem stems from one vehicle model often having different model shapes on the road. The ambiguity problem results from vehicles from different companies often sharing similar shapes. To address these two problems, a grid division scheme is proposed to separate a vehicle into several grids; different weak classifiers that are trained on these grids are then integrated to build a strong ensemble classifier. The histogram of gradient and SURF descriptors are adopted to train the weak classifiers through a support vector machine learning algorithm. Because of the rich representation power of the grid-based method and the high accuracy of vehicle detection, the ensemble classifier can accurately recognize each vehicle.

Fast stitching algorithm for moving object detection and mosaic construction

Abstract This paper proposes a novel edge-based stitching method to detect moving objects and construct mosaics from images. The method is a coarse-to-fine scheme which first estimates a good initialization of camera parameters with two complementary methods and then refines the solution through an optimization process. The two complementary methods are the edge alignment and correspondence-based approaches, respectively. The edge alignment method estimates desired image translations by checking the consistencies of edge positions between images. This method has better capabilities to overcome larger displacements and lighting variations between images. The correspondence-based approach estimates desired parameters from a set of correspondences by using a new feature extraction scheme and a new correspondence building method. The method can solve more general camera motions than the edge alignment method. Since these two methods are complementary to each other, the desired initial estimate can be obtained more robustly. After that, a Monte-Carlo style method is then proposed for integrating these two methods together. In this approach, a grid partition scheme is proposed to increase the accuracy of each try for finding the correct parameters. After that, an optimization process is then applied to refine the above initial parameters. Different from other optimization methods minimizing errors on the whole images, the proposed scheme minimizes errors only on positions of features points. Since the found initialization is very close to the exact solution and only errors on feature positions are considered, the optimization process can be achieved very quickly. Experimental results are provided to verify the superiority of the proposed method.

New automatic multi-level thresholding technique for segmentation of thermal images

A new wavelet-based automatic multi-level thresholding technique is proposed. The new technique is a generalized version of the method proposed by Olivo [1]. Olivo [1] proposed using a set of dilated wavelets to convolve with the histogram of an image. For each scale, a set of thresholds was determined automatically based on the rules he proposed. However, Olivo did not provide a systematic way to decide on an exact set of thresholds which corresponds to a specific scale that can lead to the best segmentation result. In this paper, we propose using a cost function as a guide to solve the above problem. Experimental results show that our approach can always automatically select the best scale for performance of multi-level thresholding.

A new wavelet-based edge detector via constrained optimization

This paper proposes a new wavelet-based approach to solving the edge detection problem. The proposed scheme adopts Cannys three criteria [3] as a guide to derive a wavelet-style edge filter such that the edge points of an image can be detected efficiently and accurately at different scales. Since Cannys criteria are suitable for those edge detectors that detect local extremes, the desired wavelet is, therefore, chosen to be anti-symmetric. In order to obtain sufficient information for reconstructing and analyzing the original image the dual of the desired wavelet is also required. Basically, the pair of wavelets is represented as a linear combination of translations of a scaling function. By introducing a constrained optimization process, the set of expansion coefficients of the desired wavelet and its dual as well can be determined. In order to implement the desired edge detector, a continuous wavelet has to be converted into the discrete form. For this purpose the format of the discrete wavelet transform has to be developed. Since the proposed edge filter is wavelet-based, the inherent multiresolution nature of the wavelet transform provides more flexibility on the analysis of images. Also, since an optimization process is introduced in the filter derivation process the performance of the proposed filter is better than that of Mallat-Zhongs edge detector. In real implementation, the experimental results show that the proposed approach is indeed superb.

Lane detection using directional random walks

This paper proposes a novel lane detection method for extracting various lane lines from videos using the concept of directional random walks. Two major components are included in this method, i.e., lane segmentation extraction and edge linking. At first, we define proper structure elements to extract different lane mark features from input frames using a novel morphology-based approach. Then, a novel linking technique is proposed to link all ldquodesiredrdquo lane mark features for lane line detection. The technique considers the linking process as a directional random walk which constructs a Markov probability matrix for measuring the direction relationships between lane segments. Then, from the matrix of transition probability, the correct locations of all lane lines can be decided and found from videos. Without defining any mathematical curve models, various road lane shapes and types can be well extracted from road frames even with complicated backgrounds. Experimental results show that the proposed scheme is powerful in lane detection.