Face Recognition Algorithm Based On Gabor Wavelet Coefficients Fusion

Face recognition is an important subject in artificial intelligence field. It can be widely applied in records management systems, security verification systems, credit card verification, criminal identity recognition, monitoring in bank and customhouse, human-computer interaction, etc, therefore it has gained extensive attention from lots of researchers .There are many methods proposed in the past decade years, however, their recognition precision and speed in practical applications still cannot satisfy the expectant demands of people, especially under the condition that variations of illumination, photographing azimuth or other disturbance exist in the image. The original face image captured by the recognition system usually is denoted the grey values of grid pixels. Isolated grey values of pixels cannot reflect the characteristics contained in human face directly, mapping them into the feature space to recognize through adopting appropriate transform is an effective approach. Due to the good characteristic of Gabor wavelet feature, which can link the Pixels in an adjacent region together and reflect the changes of the grey values of pixels in a local area of an image from different frequency scales and orientations, face recognition based on Gabor wavelet transform become a very popular method. This dissertation researches into the theory and technology of face recognition through Gabor wavelet transform, and we propose a new method called face recognition algorithm based on Gabor wavelet coefficient fusion by combining with image fusion technology against the redundancy of Gabor coefficients.In the dissertation, our study is mainly: (1) We study wavelet transform theory and image fusion based on wavelet transform method. We get a series of low-frequency which contains the main features of the image and high-frequency coefficient which contains the details of the image information after decomposing image by wavelet. In response to that, we use the weighted average method in low-frequency region and use the method based on region in the high-frequency region. (2) We study Gabor wavelet transform and its characteristics. Gabor wavelet transform is realized by computing the convolutions of tow-dimensional Gabor filters and the grey values of pixels in an area around a given position in an image. Gabor wavelet seems to be a good approximation to the receptive fields of the simple cells in the visual cortex of mammalians, as well as its better robustness to changes in facial pose and illumination has, so that it can get better the feature extraction of face. And then study the Gabor filter parameter choice and its significance. (3) We study elastic graph matching algorithm as well as elastic bunch graph matching algorithm. Elastic matching algorithm use two-dimensional grid to express face images, in which the node are described by a group of Gabor wavelet transform coefficients, so that a grid comparison instead a comparison between the images. The main disadvantages of this method are a large amount of the calculation, information redundancy of image, and a large storage. Elastic bunch graph matching algorithm uses facial feature points to express face, and the facial feature points are directly positioned in areas of interest, so that it reduces the redundancy of image information. However, because of the redundant of Gabor wavelet coefficients, this method still exists computing complexity and the problem of computing capacity. (4) Against the disadvantage of Gabor wavelet coefficients for the existence of redundancy in this, we propose face recognition algorithm based on Gabor wavelet coefficient fusion by combining image fusion algorithm. From the simulation results of the experiment, we get that the recognition precision and speed of the method based on Gabor coefficients is better than the method using non-fusion in a certain extent.