Dna Computing Application To Face Recognition

With the rapid development of electronics computer technology, the CPU chip of computer will come to its physical limit and so does the computer's computing ability. Because of being restricted by the computing capability, face recognition, which needs much computation, can not properly apply to large scale face recognition nowadays. By contrast, DNA computation can do much computing and parallel computing with fast speed. If DNA computation is applied to face recognition, it will satisfy the amount of face recognition's computing so that processing of face recognition can be faster and its scale can be larger.This thesis proposes applying DNA computation to face recognition. It uses powerful computing of DNA computation in two ways. One way is to quickly shrink the large scale of face recognition to small scale so that common and mature methods can be operated in it. The other way is to optimize a common method quickly in order to reduce the need of computing and raise the speed of recognition. The one of new methods, which thesis proposes, is to combine DNA computation with singular value decomposition (SVD) and the other method is to combine DNA computation with Fourier transform to meet the requirement. A new method uses the sticker model as DNA computing model, and the other method also combines with genetic algorithm because it suits to the DNA computation. DNA coding is specially designed for different situations. After doing simulation experiments with ORL face database and Yale face database by programming under matlab2007a, the results indicate that the method of combining DNA computation with SVD shrinks the scale of recognition to nearly 10 percents and still has efficient recognition rate. The results also indicate that the method of combining DNA computation with Fourier transform gets high recognition rate and it reduces the computing and raises the speed of recognition. The results of experiments show that the methods proposed in this thesis are valid. Compared with common methods, the methods show the characters of big amount of fast computing. They meet the requirement of design. The shortcoming of method which combines DNA computation with SVD is not high enough recognition rate. The other method's shortcoming is that amount of computing is not few enough. They can not be applied, because it is restricted by the physical implementation of DNA computing nowadays.This thesis presents successful experiments to prove the validity of the new methods which apply DNA computation to face recognition. It gives experience and makes some works for the future research.