Study On Image Quality Assessment Based On Eeg

With the development of economy and high technology, digital image, as an important information carrier, has been widely used in daily life. The requirement of image quality is increasing. Image quality assessment has gained widely attention, and many studies have been performed. Subjective assessment and objective assessment are the most traditional methods in assessing image quality, while they have both advantages and disadvantages. The assessment method based on EEG/ERP measurement can obtain a result matched well with the perception. The relation between EEG/ERP and image quality has been investigated in this paper.The influence of Color Saturation (CS), White Level (WL) and Contour Rendering (CR) on image quality has been studied with both subjective evaluation and EEG/ERP measurement.The results showed that the influence of CS, WL and CR on the amplitude and latency of P300 is significant. The changes of images quality due to different CS, WL and CR elicit similar responses (P300) in the brain, measurable with ERPs. The amplitude of the P300 increases exponent with the decrease of image attributes CS, WL and CR and the latency exponent decreases at the same time. Besides, the results also showed that with the decrease of image quality, the brain electrical changes tend to be restrained, that is to say, with the decrease of image quality, the extent of change will be slighter or even no longer change. The corresponding predicting models of amplitude and latency with the functions of CS, WL and CR have been established based on the experimental results in this paper.The synthetical changes of CS and WL were also studied in this thesis. The results showed that the interaction of CS and WL changes on the amplitude and latency of P300 is significant too. Based on the experimental results, a model to predict the amplitude and latency of P300 while influenced by modified CS and WL simultaneously has also been setup.The results also indicated that the subjective evaluation result and EEG/ERP measurement has a good correlation. It means that the changes of EEG/ERP are consistent with the subjevtive perception.Finally, we expanded our research to the field of 3D image quality assessment. The influence of crosstalk on image quality has been investigated. The results showed that the amplitude of the P300 increases and the latency decreases as the corsstalk increasing. It also proved that the EEG/ERP measurement is an effective method to evaluate the image quality.