Aerial Image Quality Evaluation Based On Image Power Spectra

With the improvement and development of aerial photograph, more and more aerial researchers pay attention to the quality evaluation of aerial image which is produced during aerophotograph. The distinctiveness of quality evaluation about aerial image is that many aerial images are no-referenced. Designing objective No-Reference (NR) quality measurement algorithms is a very difficult task, therefore, only a limited number of methods have been proposed in the literature for objective NR quality assessment. By analyzing the factors which make aerial image quality reduce, the method using power spectra to detect image quality is put forward. This method is effective by combining the human visual system(HVS).Firstly, the history and typical methods of aerial image evaluation techniques are introduced. Then the developing trend is pointed out, that is, integrating HVS into objective image quality evaluation is the future direction.According to the exterior operational environment and inherent principles of aerial camera, the main factors arosing image quality descending is analyzed. Then, the image power spectra is introduced and acts as a main tool to detect those factors affecting image quality. The characters are researched about image power spectra polluted by some factor metioned above respectively.To avoid the adverse impact on the result using power spectra evaluation which the noise produces, whether the image is polluted by noise or not is detected using power spectra. Because wavelet transform has the character of multi-resolution analysis and is preferable to detect singular point, we eliminate the noise using wavelet and the processed image has preferable visual effect.Discussing the typical characteristics of HVS and appling visual weighed processing to image power spectra using CSF, thereby, we set up the aerial image quality evaluation function based on logarithmic power spectra and HVS. The performance of the image quality measure is assessed by compare the IQ value to the visual image quality.Finally, the detection and estimation of simple factors which affected image quality is made in frequency-domain.The main aim in this paper is to analyze and assess aerial image quality objectively. In terms of experiment results, the method based on power spectra and...