| PART ONE NEW THEORY AND TECHNIQUE AND REALIZATION OF SINGLE-EXTRACTION FOR VISUALEVOKED POTENTIALVisual evoked potential (VEP) divided into the flash visual evoked potential (FVEP) and the pattern reversal visual potential (PVEP) is the electrical activity of cortex responsible to visual stimulation, which is immersed in the strong background electroencephalo-signal (EES). And it is extracted by superposition averaging technique in the clinic now. But it is defective that the main theoretical basis of superposition averaging technique used in extracting evoked potential (EP), so we all are seeking the new ways to singly extract EP. Based on the own characteristics of EP and the analysis of present extraction techniques for EP, We proposed the auto-reference, auto-correlation and adaptation interference cancellation technique which is called "3A" technique for short to be used to singly extract EP.On the bases of the earlier work, the theories and techniques based on "3A" were improved continuously in this research, including the theoretical derivation of cancellation factor W, the design and the realization of the screen pattern simulator of the entire visual field by self-programming, and the foundation of the novel system for singly extracting VEP, in which stimulation and sampling are really parallel. The system has a powerful performances that not only the results of the superposition averaging and single-extraction of flash VEP (FVEP) and the entire visual field pattern reversal VEP (PVEP) could be given, but the means and standard deviation of singly extracting EP could be also given, and even more result of the superposition averaging based on the single extraction could be given. Besides the function used for extracting EP, there are powerful analysis function in the system, for example, power spectral analysis, coherence analysis in time domain, coherence analysis in frequency domain based on our proposing coherence theories in frequency domain, filtering in time and frequency domain, and so on. The selected study results will be reported as follows.Firstly, we report the study results of PVEP about the young group consisting of 39 young individuals, aged 23.23±4.08 years old, including 11 male and 28 female, and the old group consisting of 44 old individuals aged 71.52±6.17, including 21 male and 23 female.The latency normal values of Nl-Pm-N2 composite wave of the young group are 99.77±13.44ms, 132.85±15.06ms, 164.96±16.06ms respectively, and the old group' are 104.95±13.11ms, 138.93±11.69ms, 173.38±11.48ms respectively. The latency of the old group is longer than the young's, and the difference is significant statistically.Secondly, we report the study results of FVEP about the young group consisting of 35 young individuals, aged 23.31±4.22 years old, including 9 male and 26 female, and the old grounp consisting of 45 old individuals aged 70.53±6.77years old, including 21 male and 24 female.The latency normal values of N75, P100, N125 composite wave of the young grounp are 87.47±19.13ms, 1124.24±19.73ms, 170.14±22.03ms respectively, and the old group' are 90.28±11.77ms, 126.28±16.56ms, 173.95±20.08ms respectively. Although according to the means, the latency of the old group is longer than the young group, the difference is not significant statistically.In this work, the frequency domain coherence analyses of single-extraction PVEP and FVEP from the symmetry visual areas were researched using the new Magnitude Coherence Function (MCF) we proposed. It is found that the coherence spectral lines of the same-type of MCF (SMCF) are generally higher than the different-type of MCF (DMCF) for the PVEP and FVEP. It shows that the coherence of same-type components in the same frequency is better in the single-extraction PVEP and FVEP from the symmetry visual area, and it has stronger inphase.The innovations of this work are as follows.(1) As the theoretical calculation of cancellation factor W is given, so the system of auto-reference, auto-coherence and adaptive interference cancellation theory used in single extraction of EP is improved, and the better cancellation of background EES could be realized, and the background EES could be canceled from EP of the single extraction.(2) Based on the AD transform technique of single-chip, the partly parallel processing of stimulation and sampling is implemented to make that the short latency components of EP can be obtained.(3) The traditional superposition averaging results and the averaging results based on the single extraction are all given. By compared between them, we can get that the traditional averaging results contain more EES.4) The post-smoothing techniques of time and frequency domains for single extraction VEP is realized to further eliminate the interference in EP.(5) It is possible that measure the amplitude variation of EP, especially the latency variation of EP each among stimulations.(6) The new parameters: the mean and standard deviation of VEP are provided for the clinical application, which the averaging technique cannot give.(7) The coherence of frequency component of single extraction VEP from the symmetry visual area is analyzed using the new frequency domain MCF theory we proposed. PART TWOTHE MEASUREMENT AND THE MODEL OFSIMULTANEOUS CONTRAST RESOLUTION OF HUMANVISION AND THE THEORIES AND THE TECHNIQUESIN THE LOWER LAYER IMAGE MINING WITH THEIRAPPLICATIONThe work investigates such a field that is from the psycho-physical foundation of human vision, related to lower layer image mining (LLIM), that is, from the relationship between the simultaneous contract resolution (SCR) of human vision and the background gray level/chroma level to the theories and techniques and applications of LLIM. SCR of human vision is also called the just noticeable difference (JND) or the minimum resolvable contras (MRC) or the contrast resolution limitation (CRL). The term of contrast is translated into "fancha (反差)" some times.We firstly reviewed certain notations about physical contrast, and then designed the measuring technique to measure the function relationship between JND and background gray/chroma level by a computer image processing technique, established a mathematical models based on the measuring data. We considered that the gray/chroma JND of human vision is the psycho-physical foundation that the existence of LLIM relies on it. For doing LLIM, we built a visually nonlinear transformation theory for gradually flattening a gray/chroma spectrum to mine the gray/chroma information of a lower layer image in order to provide visual gray/chroma information to LLIM. And then we established the Zadeh-X transformation theory which can be used to implement a lower layer image mining and hiding and image bivaluation, and carried out a lower layer image mining and hiding. The theory and technique could be as a post-processing to improve the performances of controlling, anti-body hiding, anti-stealth and low-light-level night vision. Lastly, we applied the LLIM theory and technique to stream media image mining and obtained a wonderful outcome initially.The innovations in the work lie in the following items:(1) By means of the computer image processing technique, designing a method to measure a function relationship between JND and the background gray/chroma level and building its mathematical model based on the measuring dat;(2) Proposing a visually nonlinear transformation theory for gradually flattening a gray/chroma spectrum to mine the gray/chroma information of a lower layer image and realizing it technically;(3) Building a Zadeh-transformation theory could be used in lower layer image mining and hiding and bi-valuating, and carrying out the lower layer image mining and hiding; (4) Realizing initially use of LLIM theory and technique to stream media LLIM, and founding a foundation for applying LLIM theory and technique been as a post-processing technique to controlling, anti-body-hiding, anti-stealth and low-light-level night vision to improve their performance. |
PDF Full Text Request