Y.Y. Tang’s research while affiliated with Concordia College and other places

Class specified-based feature extraction method becomes an important reference in classification, such as common vector(CV) algorithm and class frequency analysis(CFA). CV algorithm does not consider the relationship among the different classes, but CFA does while only works on the frequency domain. Furthermore, we found that the relationship among different classes is also important in recognition task. To overcome the disadvantages mentioned above, we proposed an algorithm to adjust the relationship among subspaces of different classes, and at last achieved smaller within-class scatter in each class, larger between-class scatter and closer mean centers in the remaining classes. In the case, features of each class and the relationship with the remaining classes can be shown clearly in its the whole feature subspace. Moreover, it relieves the restriction in a same subspace. We also study the comparable problem in multiple subspaces and reorganize different subspaces in our objective function. The proposed algorithm is an important supplement to CFA and CV in linear discriminant relationship analysis. The experimental results show its advantages in orl and Extend yale B face database. Furthermore, the proposed method is not limited to face recognition, also can be extended to other image-based object recognition.

This paper describes a novel method for edge feature detection of document images based on wavelet decomposition and reconstruction. By applying the wavelet decomposition technique, a document image becomes a wavelet representation, i.e. the image is decomposed into a set of wavelet approximation coefficients and wavelet detail coefficients. Discarding wavelet approximation, the edge extraction is implemented by means of the wavelet reconstruction technique. In consideration of the mutual frequency, overlapping will occur between wavelet approximation and wavelet details, a multiresolution-edge extraction with respect to an iterative reconstruction procedure is developed to ameliorate the quality of the reconstructed edges in this case. A novel combination of this multiresolution-edge results in clear final edges of the document images. This multi-resolution reconstruction procedure follows a coarser-to-finer searching strategy. The edge feature extraction is accompanied by an energy distribution estimation from which the levels of wavelet decomposition are adaptively controlled. Compared with the scheme of wavelet transform, our method does not incur any redundant operation. Therefore, the computational time and the memory requirement are less than those in wavelet transform.

In this paper, a method is proposed for offline signature verification. It is based on a smoothness criterion. It is observed that the cursive segments of forgery signatures are generally less smooth and less natural than the genuine ones, especially for those signatures that consist of cursive graphic patterns. Two approaches are proposed to extract a smoothness feature: a crossing method and a fractal dimension method. When the proposed smoothness feature is combined with other global shape features for signature verification, satisfactory results are obtained.

In recent years, wavelets have attracted great attention in both still image compression and video coding, and several novel wavelet-based image compression algorithms have been developed so far, one of which is Shapiro's embedded zerotree wavelet (EZW) image compression algorithm. However, there are still some deficiencies in this algorithm. In this paper, after the analysis of the deficiency in EZW, a new algorithm based on quantized coefficient partitioning using morphological operation is proposed. Instead of encoding the coefficients in each subband line-by-line, regions in which most of the quantized coefficients are significant are extracted by morphological dilation and encoded first. This is followed by using zerotrees to encode the remaining space which has mostly zeros. Experimental results show that the proposed algorithm is not only superior to the EZW, but also compares favorably with the most efficient wavelet-based image compression algorithms reported so far.

This paper presents linear and bilinear shape transformations including basic transformations, analyzes their geometric properties, and provides computer algorithms. The shape transformations can be used to simplify the recognition of Roman letters, Chinese characters and other pictorial patterns by normalizing their shapes to the standard forms. Important theoretical analyses have been performed to illustrate that the linear and bilinear transformations are applicable to computer recognition of digitized patterns. A number of pictorial examples have been computed to confirm the analyses and conclusions made.

In this paper, we propose a novel singularity location method to extract the skeleton of a ribbon-like shape based on wavelet maxima. The image firstly is decomposed by a new wavelet transform; then the strict local minima is computed; finally, filter the value of wavelet minima with a predefined threshold to obtain the final skeleton. The framework of our method based on above three steps is very simple with the same computation complexity with the wavelet maxima. The experimental results show that our method is a "good" method closing to human conception.

In this paper, we propose a novel affine invariant Euclidean distance of images, named annulus Euclidean distance (AED). Unlike the traditional Euclidean distance, the AED takes into account the spatial correlation between images. Therefore, it is invariant to the scale, rotation and shift transformation for images. The method is motivated that the annulus is intuitively rotation invariant, then combined with the centralization and normalization of the images, the AED is variant to affine transformation (AT). The key advantage of this distance is that it is a simple and tractable way to measure the difference between images with some deformations. Some examples in optical character recognition (OCR) and texture retrieval are presented to show the power of the AED. Experimental results demonstrate that the AED is a simple, efficient and powerful way to measure the difference between images

In this paper, we propose a statistical scheme to judge the activity level measurement (ALM) that is based on wavelet-domain hidden Markov model (WD-HMM) and maximum likelihood (MLK). The source images are firstly decomposed by the wavelets and only the coefficients in the high frequency (HH) are utilized. Considering the shift-variance of wavelets, the merged image is obtained from the source images directly. The regions of each source image are obtained by the Hough transform (HT) and their ALM are decided by the ALM of their coefficients in HH according to MLK. Finally, two multi focus images are merged by our new framework. The fusion results show the high ability of our scheme in preserving edge information and avoiding shift-variant.