| A major feature in the Network-Information-Age is personal identity becoming digital and more hidden. At present, most of China's management use management documents, magnetic cards, IC card and password.These instruments can not be avoided forged or lost and passwords can be easily stolen or forgotten. Such trouble can be avoided when biometric identification methods are used.Security Authentication System has these points, a high technology content, the accuracy of demanding certification, a great market potential as it is widely used and so on.The research projects focus on the development of the market competitiveness of the security authentication system as a full, non-contact, concealed as a system of the three major design goals.It can be suitable for small-scale, scale, space, and all the various security applications as it can be used both in Network Systerm and stand-alone operation.In such a target as the guidance, we have designed systems based on embedded devices. In a wide variety of biological features, we selected Facial characteristics as the basis for the identification.To achieve hidden identification, personal information is stored using non-contact RF IC card . RFIC card with 64 internal binary numbers, which is special each other all over the word. Cards are equipped with 16 keys dedicated space for a password storage system; there is certain cards storage space for storing data corresponding information (face mask modulus coefficient and the feature image). In numbers, keys, characteristic coefficients, feature image quadruple safety certification to ensure that the full range of security identification.Using biometric undoubtedly need to deal with a large quantity of image data.To the acquisition of images, we must first compressed and then deal with, particularly for RFIC card storage features images, and need to ensure biological characteristics under the premise of not losing on the condition of very high compression ratio.Therefore, image compression / decompression technology is a key in all the biometric identification system security technology.The Codec has two functions: first, to depress the bmp format video images which are acquired by the camera, on the one hand to save storage space, on the other hand used to the compressed video as the back of positioning and tracking feature extraction of raw materials; second, compress a head picture with a standard size to storage in RFIC card.In order to achieve maximum compression ratio and the best image results (easily for the human eyes to distinguish and the biological equipment to extract biometric feature), the experiment uses the most advanced image compression standard H.264 agreement as a reference, as well as H. 264 comparing to the previous standards adapts better to the characteristics of embedded systems, such as Integer Transform.The research of this thesis focuses on applications. Its design is based on the H.264 standard and many new H.264 study published papers in order to draw useful improvements to further reduce the computational complexity, increase compression ratio. Some typical improvements are here: intra-optimal mode selection algorithm improvements, intra prediction mode algorithm improvements, Inter Prediction improvement of the rate-distortion optimization. In this paper, the algorithm acts on in the Visual C++ programming environment in Windows operating system.The camera filming of the section containing the figure five seconds of video which uses 352×288 bmp format images, spaces about 380 million bytes. It is clear that if storage or transfer one object would require so many bytes, it cannot be accepted even in computer, say nothing of embedded systems . Therefore it must be compressed. Experiments show that using H.264 compression algorithm, we will be able to compress this section of video to within 4 million bytes to achieve a compression ratio of about 100:1. A QCIF size of a static image (about 76 M bytes), if we adopt the background processing, can be compressed to about 1700 bytes, to achieve the compression ratio of about 40:1, and if we do not use background processing, can be compressed to 2600 bytes about achieving about 29:1 compression ratio. it is difficult to find the difference between the compressed images and original images. In the follow-up feature extraction step, compared with the direct use of bmp format images, the error rate increase will not exceed 0.5%. As an application example, the encoder also consider the design of the gray scale image processing which depressed a SQCIF (128×96) gray image (about 12 K bytes) to less than 640 bytes , which makes one-kilo-byte RFIC card also can be used in the system.For our comprehensive study of embedded security authentication system, from the overall design to specific applications are an example of innovative applications, and has great market potential. This paper does the compression work for the compression requirements of the system.In view of the characteristics of images, consulting the world's most advanced compression standards' recommendations, and reference from many experts' smart thinking of the image compression encoding ,fully consistent with the needs of Feature Recognition, both compression process and the final design compression ratio belong to the world's advanced level. |
PDF Full Text Request