Showing papers by "Kuo-Chin Fan published in 2012"

Palmprint verification using gradient maps and support vector machines

Abstract: With the urgent demand in information security, biometric feature-based verification systems have been extensively explored in many application domains However, the efficacy of existing biometric-based systems is unsatisfactory and there are still a lot of difficult problems to be solved Among many existing biometric features, palmprint has been regarded as a unique and useful biometric feature due to its stable principal lines In this paper, we proposed a new method to perform palmprint recognition We extract the gradient map of a palmprint and then verify it by a trained support vector machine (SVM) The procedure can be divided into three steps, including image preprocessing, feature extraction, and verification We used the multi-spectral palmprint database prepared by Hong Kong PolyU [14] which included 6000 palm images collected from 250 individuals to test our method The experimental results demonstrate our proposed method is reliable and efficient to verify whether the person is genuine or not

Preprocessing for Images Captured by Cameras

Abstract: Due to the rapid development of mobile devices equipped with cameras, the realization of what you get is what you see is not a dream anymore In general, texts in images often draw people’s attention due to the following reasons: semantic meanings to objects in the image (eg, the name of the book), information about the environment (eg, a traffic sign), or commercial purpose (eg, an advertisement) The mass development of mobile device with low cost cameras boosts the demand of recognizing characters in nature scenes via mobile devices such as smartphones Employing text detection algorithms along with character recognition techniques on mobile devices assists users in understanding or gathering useful information around them A useful mobile application is the translation tool Using handwriting as the input is widely used in current translation tools on smartphones However, capturing images and recognizing texts directly is more intuitive and convenient for users A translation tool with character recognition techniques recognizes texts on the road signs or restaurant menus Such application greatly helps travelers and blinds

Orthogonal Nearest Neighbor Feature Space Embedding

Abstract: In this paper, a novel manifold learning algorithm termed orthogonal nearest neighbor feature space embedding (ONNFSE) is proposed to eliminate three drawbacks of the nearest feature space embedding (NFSE) approach. The first one is an extrapolation error, a feature line passes through two far neighbor points is selected for scatter matrix calculating when the distance of a specified point to this line is small. The calculated scatter matrix could not efficiently preserve the local topological structure among samples. The incorrect selection will reduce the recognition rates. The interpolation error is similar the extrapolation one. To remedy these two problems, the nearest neighbor feature space is built in the proposed ONNFSE. The last problem should be solved is the non-orthogonal eigenvectors found by the NFSE algorithm. The modified ONNFSE algorithm generates orthogonal bases which possess the more discriminating power. Experimental results are conducted to demonstrate the effectiveness of our proposed algorithm.

Falling down detection on zebra crossing at night by thermal imager

Abstract: Falling down detection is an important application for surveillance system. In this study, a two-stage falling down detection at night based on optical flow and motion histogram image (MHI) is proposed. Based on the thermal imager, the foreground pedestrian could be perfectly extracted. In the first stage, vertical optical flow feature is used to roughly detect the falling down event, then, in the second stage, vertical optical flow hybrid MHI feature is fed into the Naive Bayes classifier to verify the falling down event. The experimental results show that the detection rate is 98.6%, which demonstrates the effectiveness of the proposed method.