Analiza wpływu stopnia kompresji obrazu cyfrowego oraz apertury skanowania na późniejszy automatyczny pomiar

• Autorzy: Ziobro J. , Kaczyński R.

Warianty tytułu

EN

Analysis of impact of compression degree of digital image and scanning aperture on automated measurements

• Języki publikacji: PL

Abstrakty

PL

Przedstawiono badania dotyczące wpływu kompresji obrazu metodą JPEG i wielkości piksela skanowania na automatyczny pomiar na zdjęciach lotniczych.

EN

Results of studies concerning two parameters important for digital photogrammetry: degree of compression of photographs scanned by JPEG method and size of scanning pixel were presented in the article. These parameters were analysed due to their impact on accuracy of semi-automatic photogrammetric measurements. Compression of the scanned aerial photographs by JPEG method is applied by companies, which produce photogrammetric hardware and software: Intergraph and recently Zeiss. Loss of image quality, dependent on degree of compression, is the essential feature of the method. Equally important is cumulation of image degradation in case of double compression - when size of image is changed or resampling is done - which often happens in digital photogrammetry. The last feature is omitted in photogrammetric publications concerning JPEG; it was discussed in this article. Range of degree of compression acceptable in digital photogrametry, which does not influence accuracy of measurements, varies according to literature quite widely: from 1:3 to 1:10. Moreover, it is given in different way: as percent of image degradation or as coefficient of scaling quantization matrix. Determination of the highest degree of image compression, acceptable for automated measurements, was the first aim of this study The second objective of the study was related to determination of impact of size of scanning pixel on accuracy of measurements done by automated (semi-automated) method. The examinations were based on accuracy analysis of automated measurements of coordinates of fiducial marks. In order to make automated measurement, the texture around the measured point on a stereopair must be similar. JPEG compression disturbs image texture and in consequence directly influences accuracy of measurements of coordinates; in extreme cases it does not allow to make measurements with this method. The determination of accuracies obtained with the use of least square matching (LSM) method was studied in further works. Assessment of accuracy of relative orientation of aerial photographs forming stereopair was assumed as a base for evaluating measurement accuracy for varying degree of image compression. Inner and relative orientation are done at the beginning, and their accuracy is of crucial importance for final products of digital photogrammetry. Photographs taken with high-quality, Zeiss LMK 3000 aerial camera black-and-white negative images of sub-urban and urban area at a scale of 1:6200 - were used for examinations. In total 3 stereopairs were utilised (6 photographs). Relative orientation was done for each stereopair, for different pixel sizes and degrees of compression. Orientation has been performed on the basis of 20 measured points, evenly distributed within stereopair, using 4x5 point grid; it ensured high level of extra observations. Measurements were done on the digital plotter Image Station 6487 interactively - operator selected point for measurement on one photo, while precise transfer of this point to the second photo and measurement of coordinates was carried out automatically, using LSM method. The second part of study dealt with impact of cumulating degradation of image texture during its double compression. This problem exists, when epipolar (normalised) images are created from scanned and compressed photographs. In this stage called stereosampling stereoscopic, nadir, distorsion and earth curvature free images, with image lines parallel to stereopair base, are formed. Epipolar image also can be JPEG compressed - but this is the double compression. Resulting in cumulation of image texture degradation, as its geometry is different from the scanned photo. Epipolar images as disk files appear in older systems. In new systems these images exist only as temporary files, owing to high efficiency of image processors, so there no need to compress them. The following parameter values can be determined on the basis of the described examinations, although the study is not fully complete (it does not cover problem of compression of colour aerial photographs): Size of scanning pixel should be between 15 and 30 um. While selecting it, the following aspects should be taken into account: resolution of original photographs, sensitivity of the method of automated measurement on pixel size, as well as further steps of elaboration of aerial photographs, e.g. pixel size for orthophotomap. Degree of compression should be selected, depending on the applied method of measurement: - for single compression and 15 - 22.5 um pixel size 1:7 maximum degree of compression can be applied; for 30 um pixel size -1:5. - for double compression of epipolar images the degree of compression should not exceed 1:3.5 (for 15 um pixel size) and 1:3 (for 30 um pixel size). Automatic transfer of point to the adjacent photo and measurement of coordinates of fiducial marks with the use of LSM method enables to achieve the following accuracies: - 0.16 of pixel size - for 15 um scanning pixel; -0 .10 of pixel size.

Słowa kluczowe

PL

zdjęcie lotnicze; piksel; kompresja obrazu; fotogrametria

EN

aerial; pixel; image compression; photogrammetry

• Wydawca: Instytut Geodezji i Kartografii
• Czasopismo: Prace Instytutu Geodezji i Kartografii
• Rocznik: 1998
• Tom: T. 45, z. 97
• Strony: 137--156
• Opis fizyczny: Bibliogr. 20 poz., il., tab.

Twórcy

autor

Ziobro J.

autor

Kaczyński R.

Bibliografia

• 1. Algami D. A. [1996]: Compression o f remotely sensed data using JPEG. Vienna: XVIII Congress ISPRS.
• 2. Carswell J.D., Kern Ph. F. [1996]: An investigation into the use o f JPEG image compression fo r digital photogrammetry: Does the compression images affect measurement accuracy? Internet - intergraph.com.
• 3. Chip special [1996]: Obrazy cyfrowe. Vogel Publishing.
• 4. Forstner W. [1995]: Matching strategies fo r point transfer. Photogrammetric Week '95, Wichmann Verlag.
• 5. Harvey P., KubicK. [1996]: Experimental study o f optimal digital mapping parameters. Vienna: XVIII Congress ISPRS.
• 6. Heipke Ch. [1996]: Automation o f interior, relative and absolute orientation. Vienna: XVIII Congress ISPRS.
• 7. Intergraph [ 1996]: The Image Station Digital Photogrammetry Work flow . Internet - intergraph.com.
• 8. Intergraph [1996]: Photoscan: Photo scanning system. Internet - intergraph, com.
• 9. Intergraph [1994] : ImageStation Digital Mensuration - user's guide.
• 10. Intergraph [1994]: ImageStation Auto Orto - user's guide.
• 11. Jaakkola J., Satjakoski T. [1996]: OEEPE research project aerotriangulation using digitized images. Final result. Vienna: XVIII Congress ISPRS.
• 12. Lammi J., Sarjakoski T. [1995]: Image compression by the JPEG algoritm. Photogrammetric Engmering & Remote Sensing Nr 10.
• 13. Lane T. [1994]: FAQ - .IPEG. Electronic mail of the independent JPEG Group, [e-mail].
• 14. Madani M. [1997]: Intergraph intregrated NT - Digital photogrammetry system. Internet - intergraph.com.
• 15. Pawelczyk P. [1996]: OIJVR - trojwymiarowa przyszłość Internetu - tendencje. Enter Nr 1.
• 16. Schenk Т. [1996]: Digital aerial triangulation. Vienna: XVIII Congress ISPRS.
• 17. Wrzaskała W. [1997]: Ujarzmić piksele - tendencje, formaty plikow. Chip Nr 3.
• 18. Zabrodzki J. i inni [1994]: Grafika komputerowa. Warszawa: WNT.
• 19. Zeiss [1997]: Photogrammetry user meeting. Oberkochen.
• 20. Ziobro J. [1997]: Analiza parametrow orientacji przy cyfrowym opracowaniu zdjąć lotniczych. Prace IGiK, z. 95.