الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
High-resolution digital images that are used in remote sensing technologies tend to be of large sizes, especially with satellite imageries that provide better than 5 m spatial resolution such as (QuickBird, IKONOS, etc). These images consume large storage space, large transmission bandwidth, and long transmission times. Therefore, the archived images require compression before storage and transmission. Generally, there are two broad categories of image compression that are lossless and lossy image compression. Lossless compression methods completely preserve the original data, while lossy compression sacrifice parts of data to achieve higher compression ratios, sometimes with no appreciable loss of image quality. Generally, previous studies usually investigate the quality of reconstructed (compressed) image, while few studies address image compression effects on the processing results such as, image classification. This research presents the effects of the most commonly used compression wavelet-based formats on the classification results of the two high-resolution QuickBird and IKONOS imageries. These formats are Joint Photographic Experts Group (JPEG2000) ISO-standard format and Multi-resolution Seamless Image Database (MrSID) Industry-standard format. The two high-resolution images cover two completely different study areas; therefore the study methodology was divided into two main cases. To the end that the impact of two wavelet-based formats was assessed on the classification results. Only one classification method was used in first data set. This method was Maximum likelihood classifier (MLC) method, while (MLC) and Artificial Neural Networks (ANNs) classification methods were used to accomplish the classification process for second data set. Moreover, the compression ratios (C.R) were another difference between two data sets, where in the first one only eight compression ratios were used, includes (C.R) from 2.5:1 to 20:1. While for the second data set, thirteen compression ratios were used, includes those (C.R) from 10:1to 100:1,200:1, 300:1, and 400:1. Furthermore, only in the first data set, another effect was investigated. This effect was the influence of the traditional resampling methods which are Nearest Neighbor (NN), Bilinear Interpolation (BI), and Cubic Convolution (CC) on the classification results. Usually, resampling process is the second step after rectification in the geo-referencing or geometric correction operations. Also, the influence of resampling methods over compressed images was investigated. This research shows that the classification accuracy which derived from MrSID compressed images is better than the classification accuracy of JPEG2000 compressed images, especially at the high compression ratios in two data sets. The first data set illustrated that the NN resampling method is the best method in which the classification accuracy that was obtained was closer to the classification accuracy of original image than any other traditional methods. Moreover, it introduced that CC method is the best in case of JPEG2000 compressed images while BI method is the best in case of MrSID, especially at high compression ratios. Besides, the second data sets revealed that the ANNs classification method introduced classification accuracy higher than MLC method in case of MrSID compressed image, at higher compression ratios, while it was the higher in case of JPEG2000 at all compression ratios. Also, the results show some recommendations for some individual classes in two data sets, to make the difference in class area between the compressed and uncompressed classified images due to the image compression is minimum.