(200) R290

,., 1

Fractal Compression

By Steven Sattel

Open File Report 97-126

U.S. GEOLOGICAL SURVEY RESTON, VA.

AUG 2 8 1997

SR LIBRARY

U.S. Department of the Interior U.S. Geological Survey National Mapping Division

Contents

Page

I. Introduction 2

II. Current USGS Compression Techniques 3

III. Fractal Compression Tests 3

IV. Fvaluation Criteria 4

V. Summary and Conclusions 10

July 15, 1997

Fractal Compression

1. Introduction

The National Mapping Division (NMD) US Geological Survey (USGS), has been evaluating and subsequently using data compression since it began producing digital orthophoto quadrangles (DOQ). The uncompressed DOQ data sets will require well in excess of 10 terabytes of storage once that data base is fully populated. Also, digital raster graphics (DRG) require a similarly large amount of storage space. When the DRG data base is complete, it will require roughly 2.3 terabytes. The NMD recognized that some form of compression was beneficial for distribution to users because of the large size of the files. The lossy compression of these large files saves space on storage media and distribution media, such as CD-ROMs, and also saves time in transmitting the files over a network. Some users find that the minimal information loss is acceptable and does not compromise their intended use of the DOQ data.

Storing and archiving the original uncompressed digital data on tapes is necessary to maintain all of the information. Because the compressed data offer a lower cost, easier distribution alternative, the NMD can reach a wider market that includes organizations that have smaller budgets, less powerful workstations, and less stringent file and image fidelity requirements than its traditional customers have.

Because of significant coverage in the technical press (see Appendix 1), the Office of the Division Chief, NMD, requested that System Development and Integration (SDI) staff investigate the feasibility of applying fractal compression technology to the digital cartographic data that the NMD must compile, store, and distribute. This study was accomplished through a contract with the Intergraph Corporation. FC Imaging (FCI), formerly known as Fed Comm International, was a subcontractor.

The Fractal Compression Study is part of a continuing effort by the NMD to investigate various compression techniques. Previous efforts in compression research by NMD's Western Mapping Center's (WMC) have yielded the following two related reports:

Data Compression Effectiveness on Digital Ortho Imagery Using_JPEG Software by Robert Vitales USGS, Menlo Park, Calif May 1993

Any use of trade, product, or firm name is for descriptive purposes only and does not imply endorsement by the US Government. 2

Compression of Digital Othophotos by John Villasenor, Ben Belzer, and Min-Jen Tsai UCLA Electrical Engineering Department September 1993

The second report compared wavelet-based compression technology to Joint Photographic Experts GroUp (JPEG) compression.

Current USGS Compression Techniques

A. Standard DRG's

The standard product is compressed in the following way:

1. For USGS - scan at 500 dots per inch (dpi), resample to 250 dpi The contractor (Land Info) scans in 250 dpi format

2. Georeference - does not affect file size

3. Standardize colors - reduce palette to 13 colors

4. Run length encoding (RLE) compression yielding a standard Tagged-Image File Format (TIFF) file.

5. Descreening

B. USGS - DOQ's

DOQ's are compressed using the JPEG image compression algorithm. The compression is applied to the final product and is also an intermediate step in thy: production process.

III. Fractal Compression Tests

The fractal compression tests were accomplished as follows:

A. Input.

DOQ's

a. Southwest Atlanta, Ga.

3

b. Felix Peak, Mont.

2. DRG's

a. Jackson. Miss. b. Output from the Product Generation (PG) software for:

(1) Arley, Mo. (2) Calvert City, Ky.

B. ;Methods.

Intergraph Corp. first converted the files to 24-bit standard Microsoft format which is known as "bmp" because the software that they used required 24-bit bmp for input. Iterated, Inc.'s, Multiprocessor Image Toolkit (MPIT) software was used for the compression.

C. Compression.

Intergraph used the following compression technique:

Iterated's MPIT is designed to take full advantage of symmetrical multiprocessor (SMP) systems (yet they aren't required for operation) and run reliably in both large-image and high-volume still image applications. On SMP/NT systems, MPIT converts large image files into a series of smaller segment panels then sends panel 1 to processor 1 , panel 2 to processor 2, and so on , (maximum of four processors) for compression. When a processor finishes compressing its assigned panel, the next panel in the image is sent to that open processor. This process continues until all panels are compressed and recompiled, resulting in the much smaller fractal image format (FIF) file. Compression ratios and time to compress are dependent on user-defined quality settings. The input file must be a 24-bit BMP file.

IV. Evaluation Criteria

A. Performance

Compression times varied according to the hardware configuration that was used (number of processors, Mb of RAM). The following compression table (table 1) shows fractal compression results for four different hardware configurations. No time information is available for

4

the current DRG compression method. The current JPEG compressions generally take less than 4 minutes on a Data General 530 using one CPU with 64 Mb of RAM.

2. The decompression table (table 2) indicates the time to open (decompress) and view each file for the various compression methods. All decompiessions were performed on the same machine with the same number of other windows open.

(See the tables on the following pages)

B. Implementation Costs

At the moment the only costs would be the Iterated Systems, Inc. fractal compression software and Intel-based machines to run the software. The cost per license for the software is $3,500 per processor. Some customization is necessary to use the software. This price is for development (customization) only, but developing requires only one processor. The software uses multiple processors when it compresses. Currently, the viewing software is available only for the Microsoft Windows environment on the Internet at no charge. Iterated has stated that they have no plans to start charging for the viewing software. Also, they have given permission for other organizations, such as the NMD, to reproduce the viewing software on distribution media, such as CD's at no cost.

1. DRG cost estimate:

Assuming an average compression time of 1 hour per DRG and 54,000 DRG's (the number of USGS Quadrangles), then

1 hr x 54,000 = 54,000 hrs / 16 hrs = 3,375 days / 48 weeks = 10.04 yrs

This assumes that a batch job could be implemented to automatically start processing the next file as soon as the previous one finishes and that the machine(s) would run 16 hours/day, 48 weeks/yr. Also, personnel need not be present while these computer jobs are running. Therefore, jobs could be run overnight and during weekends.

1110 machines with dual processors were purchased, then the entire job could be completed in a single year. Of course, the DRG data base is only partially populated; therefore, we would need fewer machines to

5

Table 1. COMPRESSION

USGS Input File Uncom- I Standard Output Ratio of Standard TD-5 / 2 P5 Intel ISMP64 / 4 P5 TD-400 / 2 P6 TD-400 / 4 P6

Quality Size (24 Bit pressed Product i File Size Product to Fractal CPUs 128MB Intel CPUs Intel CPUs Intel CPUs File Factor Color) File Size File Size (FIF) Compression (x:1) Memory 256MB Memory 64MB Memory 64MB Memory

DRG Calvert City, Ky 106.53 35.51 6.43 1 30 1 1.7001

! 3.78 (1:04:51) 00:28:12 (0:28.36) (0:14:45)

50

70

4.100 i6.400 '

1.57 (1:37:12)

1.00 01:59:09

00:42:20

00:51:55

00:43:57 (0:23:00)

00:52:01 00:27:12 DRG Arley, Mo 106.53 35.51 3.961

30 1 1.200 3.30 (0:50:22) 00:21:54 00:21:33 (0:11:46) 50 j 2.700 1.47 (1:12:00) 00:31:23 00:32:01 (0:16:03)

70 4.200 0.94 (1:31:16) 00:39:41 00:38:10 00:20:19 DOQ Felix Peak, Mt , 1 128.7 , 42.97 5.50

30 1 1.100 5.00 (1:31:51) 00:39:56 00:40:09 (0:21:00) 50 3.000 1.83 (1:23:50) 00:36:27 00:36:49 (0:19:06) 70 ' 5.100 1.08 f1:31:51) 00:39:56 00:40:09 00:21:09 ,

DOQ SW Atlanta, Ga 151.7 50.62 ' 5.57

30 1.100 5.06 (1:06:58) 00:29:07 (0:30:00) (0:15:10)

50 2.800 1.99 (1:51:33) 00:48:30 (0:49:39) (0:25:22) 70 5.600 0.99 (2:15:10) 00:58:46 01:00:18 00:30:42

Notes: Items in () are calculated (projected) times (h.mm.ss) and are left justified; All actual times are right justified; Compressica Compression rates in hh:mm:ss/Mb; All files sizes in Mb.

FRACCOM3 WK4 07/16/97 03 14 PM

Table 2. Decompression (Load Times) Corn-

pressed Time Ratio Input File File Size Compression Time to Standard

File Size (Mb) (Mb) Ratio (x:1) hh:mm:ss I'roduct (x:I)

DRG's

Jackson, Miss. (BMP)

Standard Jackson, Miss.

Jackson, Miss. (FIF)

42.12

42.12 11.540

42.12 7.330

3.65

5.75

00:03:17

00:03:53

00:00:21

0.845

0.090

PG DRG's (BMP)

Calvert City, Ky.

Arley, Mo.

35.51

35.51

00:02:48

00:02:29

0.757

0.851

PG DRG's (standard)

Calvert City, Ky.

Arley, Mo.

35.51

35.51

6.430

3.960

5.52

8.97

00:03:42

00:02:55

PG DRG's (FIF)

Calvert City, Ky.

Arley, Mo.

. 35.51_ 6.394

35.51 4.210

5.55

8.43

00:00:15

00:00:17

0.068

0.097,

DOQ's (uncompressed)

Felix Peak, Mont.

SW Atlanta, Ga.

.,

42.97

50.62

00:05:47

00:06:44

1.483

1491 ,

DOQ's (JPEG-standard) _

Felix Peak, Mont.

SW Atlanta, Ga.

42.97

50.62

5.500

5.571

7.81

9.09

00:03:54

00:04:31

DOQ's (FIF)

Felix Peak, Mont.

SW Atlanta, Ga. _

42.97

50.62

6.120

5.648

7.02

8.96

00:00:13

00:00:12 _

0.056

0.044 Notes: The column labeled "Time Ratio to Standard Product" is the ratio of the time to decompress that file to the time to decompress the standard product (JPEG for the DOQ's). "FIF" means fractal image file. "Standard" means the standard USGS product. "PG" means the output from the product generation software. "BMP" files are the uncompressed DRG's.

catch up in a single year. The exact number of machines depends on the percentage of DRG's that are complete when this compression technique is implemented. If more than two processors were purchased for each machine then the task could be shortened further. Each machine would cost about $10,000. Another implementation cost would be the cost of a programmer for minor software development because the software requires some tuning to fit the individual circumstances.

The total implementation costs would be as follows:

Hardware $100,000 Programming $10,000 ($50/hr x 200 hrs.) License $3,500 (*)

Total $113,500

* Only one license is necessary for development because executable modules can be distributed without individual licenses.

Realistically, only one machine may be necessary because the DRG's cannot be produced nearly as fast as they can be compressed.

2. DOQ cost estimate:

Assuming an average compression time of 1 hour per DOQ and 216,000 DOQ's (the number of USGS quarter quadrangles), then

1 hr x 216,000 216,000 hrs / 16 hrs = 13,500 days / 48 weeks = 40.18 yrs

This assumes that a batch job could be implemented to automatically start processing the next file as soon as the previous one finishes and that the machine(s) would run 16 hours/day, 48 weeks/yr. Also, personnel need not be present while these computer jobs are running. Therefore, jobs could be run overnight and during weekends.

If 40 machines having only dual processors were purchased, the entire job could be completed in a single year. Of course, the DOQ data base is only partially populated; therefore, we would need fewer machines to catch up in a single year. The exact number of machines depends on the percentage of DOQ's that are complete when this compression technique is implemented. If more than two processors were purchased for each

machine, the task could be shortened further. Each machine would cost about $10,000. Another implementation cost would be the salary of a programmer for minor software development. A programmer is needed because the software requires some tuning to fit the individual circumstances.

The total implementation costs would be :

Hardware $400,000 Programming $10,000 ($50/hr x 200 hrs.) License $3,500 (*)

Total $413,500

* Only one license is necessary for development because executable modules may be distributed without individual licenses.

Realistically, only one machine may be necessary because the DOQ's cannot be produced nearly as fast as they can be compressed.

C. Benefits

There are three potential benefits for using fractal compression. They are:

1. Smaller file sizes would reduce storage space and transmission times. Appendix 2 has information about an Internet demonstration related to this topic.

2. The fractal compressions appear to have better quality and accuracy than other techniques give (less loss). The fractal compressions appear to have less data loss, particularly in the case of the DRG's. More detailed and specifically quantified investigations are necessary in this area (see sec. V). A CD with sample compressions is being distributed with this report. The following is a list of criteria that should be used to judge fidelity and accuracy:

a. Does the compression scheme move edges? How far? b. How much is their relative separation changed on a scan line-to-

scan line and pixel-to-pixel basis? Dr. Villasenor's report (Compression of Digital Orthophotos) includes some excellent graphs that address these issues for discrete wavelet transform which is another compression technique.

7

c. The NMD should analyze the JPEG performance and use those numbers as indicators of the maximum acceptable ranges for these types of variations.

3. Faster decompression. The fractal compression of the Jackson, Miss., DRG took 21 seconds to load in the viewer. Its file size was 7,329,395 bytes. The standard DRG product for the same quad on the same machine with the same number of windows open took 3 minutes and 53 seconds to load in Adobe Photoshop. It had a file size of 11,538,508 bytes. The difference in performance while panning and zooming was similar.

D. Disadvantages

1. One of the major drawbacks of implementing fractal compression at this time is that the Commercial Off the Shelf (COTS) software only works on an Intel-based machine running some version of Microsoft Windows. Most user organizations have some of these machines, but any organization that does not would have to purchase one. This could be a hardship. The vendor has discussed the possibility of porting the software to other operating systems or platforms, but so far this has not been done.

2. Another disadvantage would be that the NMD would have to distribute the viewer with the data, free of charge. There is no monetary cost for NMD associated with it.

3 Currently, the free viewer is not powerful enough to be useful for most USGS customers. It can zoom in, but not out. It cannot pan, and it cannot generate files in different formats, such as JPEG or TIFF. The last deficiency would preclude overlaying images stored in a different format with images stored in the fractal format.

4. At this time, there are no widely accepted, commercially available applications. For instance, Adobe Photoshop does not accept the fractal format, even though it does accept most others.

Fractal compression lacks any national or international standards (see sec. E below).

5

E. Compression Software Maturity

1

2

There are no standards for this compression technique. However, only one company is defining the FIF file format. Standards do exist for the compression techniques currently in use for the NMD standard products.

The software is not available on most platforms (see above).

3. The system requirements for the Fractal Imager software are as follow:

a. Shown below is the minimum configuration: Intel-based PC 486/33 or higher. Windows 95, Windows NT, or Windows 3.X. (Windows 3.X requires Win32's version 1.30.166, which can be installed from the Fractal Imager Web site or from a CD.) 8 Mb RAM. Windows-compatible VGA or SVGA card.

b. To get the best performance from Fractal Imager, you should use the following configuration, in addition to the requirements listed above:

20 Mb free disk space for handling images. A 24-bit, true-color video card. 16 Mb RAM.

What are the impacts of these system requirements? Undoubtedly, the USGS would lose some customers. Would it be an acceptable number of lost customers?

4. There are other organizations using fractal compression technology. For instance, Intel Corp. has fractal compressed images at their World Wide Web (WWW) site. Copies of their web pages are in appendix 1. Also, MCI, in cooperation with the USGS, has a fractal demonstration on the WWW, and copies of these pages are in appendix 2. For more information on fractal compression technology and links to other fractal compressed galleries, visit the Iterated Systems, Inc., home page at URL http://www.iterated.com/.

Most of the companies that are primarily using fractal compression do so to save transmission time on networks, including the WWW, and to save storage space. MCI is currently studying the possibility of using fractal compression to transmit videos on demand. This idea can only be implemented if transmission times are small enough.

V. Summary and Conclusions

The study indicates several advantages of compressing image data in general and, more specifically advantages of using fractal compression technology for that purpose. The advantages of compression are that it saves space on both in-house storage media and distribution media such as CD-ROM's as well as saving time when transmitting the files over a network. Figures 1 through 14 at the end of this report are hardcopy reproductions of each compression method. Figure 3 is a 600 dpi scan of part of the Jackson, Miss. quadrangle. Figure 4 is the fractal compression of the 600 dpi scan. Notice that the quality of figure 4 is better than the quality of figure 1, which is the standard USGS product. Also, the fractal compression uses only 80.6 percent as much storage space. For the DOQ's, compare figures 10 and 13 (JPEG compressed) with figures 11 and 14 (fractal compressed). In the case of the DOQ's, the quality appears slightly better for the JPEG in roughly the same amount of space. The findings indicate that fractal compression can slightly decrease file sizes while giving as good or better image quality especially for the DRG's. Another important advantage is that it takes much less time to load the image into a viewer. The fractal compressions of the DRG's can be loaded in 8.5 percent of the time needed to load the standard product, on average. The DOQ's can be loaded in 5.0 percent of the time, on average. See table 2 for the specific differences in load times.

The author believes that fractal technology is not sufficiently perfected for use in our production processes and that the NMD program managers should monitor the further

development of the fractal, wavelet, JPEG, and other compression technologies for incorporation. This study shows that the greatest potential benefit from using of fractals would be in the quality of the DRG's achieved by using an initial scan with more dpi while using fractal compression to limit the file sizes. Additionally, fractal compression has the potential to become a Division standard for all of our image data because it works equally well for both DRG's and DOQ's. At this time there is no standard format. JPEG is used to compress the DOQ's and RLE is used for DRG's because JPEG was found to be unsuitable for DRG compression. A standard format would provide the convenience of overlaying different types of images without the extra step of converting to a common format.

The following pages have printed examples of the image files. The images have been enlarged by a factor of roughly 3. This shows artifacts (patterns or marks that are a degradation of the original image) from compression better, and it minimizes the effects of the low-resolution printer that was used. It should be noted that the printer adds an additional amount of artifacting that may not be present in CRT viewing.

10

..

1

1

• 11•••••••••71 t;"--

...•„„v, •

•

"7"----"--716• ""K-att

4, WI t .r

• -iet:44)

,..

e.listrAgn

• .. ,

I. , .,

I

r .1 , 16.

,t E ', 1

.",•- ''' .-.. t. ii %

1,0. r . . -1/.., .,..•-.. . .

N, ....• '

..-.. -, , .. ,

fi-, 1 i • - -.. - . ...-. c '. 4,-••• 0 4,. ?..i k 1i....______10....,E.2.......Y.

v.„.. ‘..,.... „..

4,,,2. -''.~•-• • ..,0-te-1--......„,„ ,- .1.... ,........_...._, --1-7,

;?. 1 1;i

i P i ..•.,........,,....--'...........,,„,.......•.....,EMII.,Mv•..00.r.r-.... .11;.

..........,-...,-.........., • um- -vir . -.„ ti.

I ' . i 1

;•

'2 i CS !

ft1 4.,,_.....:_....."-•,•-4....7-4,"—'-"""""-'11,0 .4,,,_ :, 1

•

r 1 I i• _,

. .-4-.1--. ••••,. ''',, / • -`".•—•

..o..d•Nmsr....L....mmmos•yomo.....00.m•moso.amr•maoo•Jr--•'• • .- lakibs Self. "--.• - „, •

L4E2 - 3 4 4 •

ii 4.; C - •

4

T-7-1 t ii^ • _ 1 .t 1 i14 4--- 1 era J•inw ...NA. 4

Compression Method - ORIGINAL USGS INPUT FILE (RLE Compressed) File Name - JCKSNMSG.TIF Source of File - Standard USGS Digital Raster Graphic Product Compression Ratio with Input File - N/A Compression Ratio with USGS Standard Product - N/A 3ytes/Square Inch - 19,425

Figure 1 ArD k (;()I.I)0C

41, O

- . -• N.

•

I

• t:•

11 .) • \"I?*14 1

•

•, `s-,1 /4

.!

44( N t\

•

41,•

•

V•••.,,M, •

• crow-7 -e•••=•••

..•••••••.a.

1 ?‘,

gt••

I

Sch 4,••••••

•

4 ,,.... ....................ti,k,ge..--- t.--_-.•-,•-•,, T---'--••

3 4 ': C

ii1M

, ,,A • ill • i . ..t•—.----”."-

...„ -,. ..-,

I ---..---.

! I mat...•:-.2•0•-•••rwor

-Neu

1 4 4

Itt •

Compression Method - Fractal Multiprocessor Image Toolkit (at high quality factor - 70) Input File - JCKSNMSG.TiF Source of Original Input File - Standard USGS Digitai :Raster Graphic Product Compression Ratio with input File - 1.58:1 Compression Ratio with USGS Standard Product - 1.58:1 Bytes/Square inch - 12,339

Figure 2 JAC DRG05.DOC

W-•

• .{

.--I•..k'' ,---•::::--..--, , . , .,'..:.--.-;,."------ \

4"1.: x rr. 0 “ \ • \ \ . '-, •••-•'....,•'** -s--..\\ —r, .... 4,1 % ''''''‘‘. \

.d•-"'".0 \\ r ,*, \',.,, -,.., -,......"",,.**>,. ....,.. '`......____,•'

"t '-`",,

*4:

S' c

ion •

,r6

r.

• •Y,

Compression Method - N/A ORIGINAL INPUT FILE File Name - JKMS600.TIF Source of File - 600 dpi scan of the Jackson, MS quad Compression Ratio with Input File - N/A Compression Ratio with USGS Standard Product - N/A

Bytes/Square inch - 871.831

Figure 3 JACI)k( )(:

.

• • 2 ••••,••••••••••••••••7. ";•••••ri4

NN.

gib.N:S:\

f

• , f- rA

N

S

oe c''' . ,,..........,

/ ,3, , ./....., , '

Vi dlr.' .N.,.

•,...`•••,...., eit .,..)AT) ....,'"' ......0. .4.....—

4.........(4......, ..., •• • - • •••0• -.......,,ir,...L. ,),,,,. ..."..,...

• '-‘ •-,

*, \ As li*N. \,,,,...,4,-:...,...,.._ -T.,.... 2.t ''',„ \‘‘ . 'ft....4.Ni ‘-'!....,,, --''''... -

i

EA ,...-•••••••••••••••

A

...... ,__. .... Thil--.‘a__:,.,..„.7121,04-iiriiirk-,-.-1- ..:.1.

I I1111111111111111

_='7-

i. !

•••••—••••••••. ,"•"*"""''' "' • L

x , e, -•-.4.

'....is,"—...s............/ i

.04."4.' 1 , '

' • ..•..i1*

Compression Method - Fractal Imager input File - JKMS600.TIF Source of Original Input File - 600 dpi scan of the Jackson. MS quad Compression Ratio with Input File - 55.7:1 Compression Ratio with USGS Standard Product - 1.24:1 Bytes/Square Inch - 15.653

Figure 4 MCI )1“2.08. DOC

•

N. -

•

_,--..--"-..—.—........--7.....,*,.. ......-

-;-: -,---- --- -..._. . .---‘-..,'• '...

/

_ • 4,

II),HAA. . ,

'11.1

ic;?.

rl Ij II '•• ?, ;1.

I\ I\

BM \ IL\ • \s„) „<„

.

•••• •-• -...., -,.. , I

'I ,-, 4.. Al. .... : ', •‘..,,,'Is It. .....\ , . , ,, ...._. .... ., .. '

.. .... .„.

7.

,alk6 •„,......._-:.._,....._...... ‘..-/-----,_.• -.,,,::::.- ,--- is,-

,•

4--- 7.... . ' • —

Gaging tation , , •— •1”.

' . 1 1 —

-7 -Kentucl Dam

; • .:*

\Boat Ban

Compression Method - N/A Original Input File File Name - CAL_CITY.TIF (packed TIF format) Source of File - DLG from PG Compression Ratio with input File - N/A Compression Ratio with USGS Standard Product - N/A Bytes/Square Inch - 10.820

Figure 5 CALDRON. LX

• •

V' i• ; i 1-.7 ,,,-- ,- i.

.• .e•,;• • ,. , ... ale

.• ,,,,, , , ,ei ‘, .

4 - i I1 II .4"? ..K.,.. ,. •••----• I i .31 1

i ----_____, .1, i • l .a i" .•

\.,____ ",;:,::-•.:.,...\ -._ ,, tii i

.••••'

• e"--"•••••

_

- •

••• I

1 ( I , ti C:1[1— I•I -1.1 ics

t r, •—••j

n" ' . , .... ....,

1.1 1:'7/ ..,_ --,/ a , — ,,--o--7-4.-......, --,,t.

..-_,../. :....... ...." ,,

v.,,,..1 / .. 1 i ,•-.' 1 (f.------)1' ."• 1

../7.• / ii-lr'' II11,-- ///(./ ..i

''') r il \ t ; I

• .

;.--- .1,—, ',I.:::•:',. ..-. - ••• • ' I '

• , ..____ ... _....,,

. ......---•

••

,• , . , . . ... ,,..

... '1. ,„,...‘ • \ . -• i \ \ ,. ,

---, .....,..,.'".'

Walker •• —

Is, --t,

Gaging tation i'•

• r •21 330 Kentua Dam ‘ Boat Ba n1

4‘.rl*

Compression Method - Fractal Multiprocessor Image Toolkit (at high quality factor - 70)

input File - CAL_CITY.TIF (packed TV format)

Source of Original Input File - DLG from PG

Compression Ratio with Input File - 1:1

Compression Ratio with USGS Standard Product - N/A

Bytes/Square Inch - 10,764

Figure 6 CALDR(11().Ex)c

- ,•

) Boat { r n

i i

,.;

i • ,.-

't '1 i

_` \L,

I 1

i

.,.,' .NI\ AIL 7-----_:,

I

I,

r•-

'

- —

•

1 --- ---,

,t .-I,

i ..• -

‘ (.'

\ \• .,

\

'It..., Ns,„ ., •,„..

,i

....

..--,...-•

....

.........

.......--*,...:

\

o•

.. ../

'-

VILLETI EiFVATIO 807

/-

t

Compression Method - N/A Original Input File Hle Name - ARLEY_MO.TIF (packed TIF format) Source of File - DLG from PG Compression Ratio with Input File - N/A Compression Ratio with USGS Standard Product - N/A Bytes/Square Inch - 6.672

Figure 7 \momnuxx.

- - /

-**/-..4...-...-„ ,i'. ' _ _-.4-.. ...

I ..

t,„ / i )1 - ..--1 - ,.------ ! ...„

•)•1/e 1

'-'----: -f-Boat

1 Ls, „0", 41 /

I jII

. \ _ /

,-,Th. 1. ..

t 1 4 '' .1;, ,

.../

/

.171 VI L.L. 4:1 El-FTA Tiq 86, 7

-s•-•

' . (

Compression Method - Fractal Multiprocessor Image Toolkit (at high quality factor - 70) Input File - ARLEY_MO.T1F (packed TlF format) Source of Original Input File - DLG from PG Compression Ratio with Input File - 0.94:1 Compression Ratio with USGS Standard Product N/A Bytes/Square Inch - 7.088

Figure 8 .1 R1.1.)I2C, 1 2. 1 )0('

ti

011,

CO

(/)

0)

0 0

0

3 3

_C

rt w

0 0

_C

1-6

0

C

W ' C

C C

C

a)L_ .

-oc oc

-_

_

cv (I)

U)

EO

U

)c (())

CT

cu a) (/)

c)_ U)

oE _a)

0̀;)-0E

0E

a)

C) u

0D

C) C

) co

~•n~rtria•ni~~1r

• •

_

_

• . •

•• N .

;0 •

.0

. ... •

• 4̀.,•

. • t V

s'vwt'v

v.:mew

/A•

s,•---arw

tr k:'•••

A

4.. •

V

a

t0 _c O

sr

.c

0

. _

O (13-0C(13

D L

L (f)

.0

5 0

LLI W

Q (/) c

Ni

0_ u_ L i 7 C

D

(.0

--) r-.

S <

a

3 -§

T.—

0

C

,

.0 0

—

. 0 0

_c

al CC (13 —

•-;-- - o

O

c

(z"z3

c-co i_

C

Q)C

0 -

0

2 (n

oco ( n

0.(D i 0

. (,) 0'

L_CD 6=

Li)

0_2? LP. (...i.;)1,)

o_ E

5D

EE

a)

0 0

-0

0 0

-5

, ao

Compression Method - Fractal Multiprocessor Image Toolkit (at high quality factor - 70) Input File - GEORGIA.TIF Source of Original Input File - USGS Digital Orthophoto Quarter Quad Compression Ratio with Input File - 8.96:1 Compression Ratio with USGS Standard Product - 0.99:1 Bytes/Square Inch - 10,404

\I I I)( )Q21 1)()(Figure I I

73 cm =

a

<

LLJ

t_ Z

...J

w

a'. r

t5ca

D

7:3D

a

0

c 1 .. 6.:

z 0

_

15

_.J _c < -2

< Q

. a3

3

.cz 0

z 2

t

a) gEZ

0 c.L-

0

73 CD

4.,‹x LL

..... a_ u

) ,--, p

cr ) c-D

r--Z

• a

r--

_c .c

i <

CNA _ r

(/)-0

Z

en § •§

cn

o

<

---'

_c

I. (/) o o

c

i; Z

n .tc

(7; 0O

. ct cr c

w

c c

E.)

(;) ii Cl) ,(n

(., 3o

Cl)

al a

lal cncr(%

(1)Z

U CL CL

ci 0

O

oE

w

E E

C)

6= OD CD C) OD

• ••••

Rfa)t00-C

11 A - is i s t he tan dard Produc

o-Co

ct To4,

.R. ,

8. C

O

(13

V I

C0 i C

O() 1

) ("7

5D

ir- ,V

I O

1 4

5

CD

LL1

a) a_

( n tiCO

-0

EL • §

. §0..c

—

0 0. _

Fes-C CI .47"

Ill C

(13

C

• .6

, fy

CC

c

• i7._ c c

a) . _

. =

. _

(II 0

0

.00

N

s._ w

U)

u) a)

0

ti) (

./)

Cr

a) 2). i .

—

a) (f)

o_ 2

CL

CL

E

D D

E

E

a) o

O-wo

cs cs

U

c

CO

.t is •u

alit factor

0L..a)t0

cr) 01 -

cu O_c0

10Ocii-

Uo

-72 2

u)

0 ' (7 a,)

DLi=

u)

ca

LL

2 0

_

u)

<

_C

C'ec

(E)-0

<

0

_C

c6 z c(3

C

0 E

n fYC

0

. 0

0

(13 (7)-v)

o

Cr

(1) ) a

) L

L (1

) 0.

U)

a)

5E

Et

ED

0 0

-0 0 0

:(1

))C.) (n C

.) G

O

Appendix 1

come to Intel http //www Intel co

I

see what's

on the edge in

web development

c focus, F.tt open the door

to the world

withlr, your PC

Y

To receive your own reg,ularly updated content, go t 6'ews for

Table of Contents of this site This site best viewed with Netscape 2.0

ui :9t

50.144s.t based web

Trademark Information. (c) Intel Corporation, 1996

36/20/96 08 45'

• •

ii-noi;.)-,ay First http /Iwww !Mei com/techone/index h

Intel

able tr:lvtzi in iJrna 11I With<31-4t io4ving..... . . . . .

your chair ,

• see what's

2Cc MHz on the edge in Perstfurr: processor tsrirtg& new Fife to web development

nimabo

4,„.4":4 :4 4 the multipia yer adventure game that

combines 'ova and vrmt

Great images...small file sizes? New Pentium(R) Pro processor VRNIL Explore Fractal Compression and see L.-s.;:..i -iions—Now compressed for how it works! faster download

If you are a designer or developer and you are looking for information on Intel products please refer to Info for Developer:,

/1,./(A71:011,

06/20/96 08 41

7oduction to Iterated's Fractal Viewer http //www Intel com/techone/fracimg/intro h

fl te rriagp

corn ression galiery

Great images...small file sizes? A fast Pentium(R) processor and fractal compression make it possible!

Introduction

Iterated Systems' Fractal Image Viewer* is a Netscape Navigator* 2.0 plug-in that allows vou to view and manipulate imazes that have been dramatically compressed (we have encountered compression ratios exceeding 50:1) using fractal compression technologies.

A fast Pentium(R) processor lets you trade processor performance for precious network bandwidth by allowing you to quickly uncompress and manipulate these extremely compressed images.

Check it out!

To see what it's like to use fractal compression. fire up your Pentium processor and get equipped with the Iterated Systems' Fractal Image Viewer plug-in available in the tools section.

We've assembled images from- around the world of Intel in our tzallery for you to check out. Enjoy the show!

Trademark Information

:•,f 06/19/96 09 31

http //www Intel com/techone/fracimg/gallery hgage Gallery

5 r ;<•.;Z• t •' r,5;;;;f:

Don't forget to use the magnifying glass to zoom in, and the right mouse button to access other image manipulation features!

Pcntiiiin" Processor

FIF size=64.8kb FIF size=25.6kb BMP size=1.17Mb TIFF size=515kb

Resolution=1280x960 Resolution=673x469

FIF size=20kb FIF size=20kb TIFF size=1.12Mb TIFF size=816kb

Resolution=631x691 Resolution=547x561

FIF size=21.1kb FIF size=67.4kb TIFF size=708kb TIFF size=3.05Mh

Resolution=342x664 Resolution=1124x1670

06/19/96 09 43

age Gallery http //www Intel com/techone/fracimg/gallery h

FIF size=82.7kb TIFF size=1.86Mb

Resolution=1217x787

ft; door t..• iantiN •

FIF size=97.7kb TIFF size=1.90Mb

Resolution=1211x792

FIF size=28.5kb TIFF size=712kb

Resolution=-157x685 Can you spot our CEO, Dr. Andrew S. Grove, in this 1968 shot?

Tracleinark

06/19/96 09 43

Appendix 2

Fractal Demonstration http //xenon proxima corn 8080/us

ractal Demonstration

South Florida Ecosystem Program Ecosystem project with fractal map

Fractals Information on Fractals

and Fractal Viewing Software

SPOnSartNiby mcg

United States Geological Survey

WWW Home Page

U S

USGS Catalog Online ordering of

USGS Mapping Products

06/20/96 08 47!

--enstration of Fractal Compression Technology http //xenon proxima corn 8080/usgsiinfoidemo r

Information on Fractals Benefits of Fractals Fractal Viewer Software Fractal Gallery

Fractal Compression vs. GIF or JPEG compression

Fractal Compression offers better compression than either GIF or JPEG Images that are compressed with Fractal algorithms are better suited for an environment where size and download times are an issue. The table below lists the file sizes of a selection of images in different file formats. If you wish to compare image quality four formats are available for you to download.

How to View these files -

Targa (TGA): Download the image to local disk and view the image with another application. Try LView if you do not already have a viewer.

Your browser should support GIF images. If you would like to view them outside your browser, download the image to local disk and view the image with another application. Try LView if you do not already have a viewer. JPEG: If your browser does not support JPEG images, then download the image to local disk and view the image with another application. Try LView if you do not already have a viewer. Fractal Image Format (FIF): Download the Viewer and configure it as a helper application for your browser.

Click on the file si:e to view

06/20/96 08 52

,,-nonstration of Fractal Compression Technology http.//xenon proxima corn 8080/usgs/info/demo h

Description Targa (TGA) GIF JPEG Fractal (FIF)

3428k 513k 225k 151k

8702k 1877k 634k 422k

2904k 487k 95k 141k

5164k 553k 217k 169k

Zooming Capabilities Fractals are infinitely complex. Zoominu, in on a Fractally Compressed image just requires some recalculation on the client end so that there is no need to download another image to get a closer view. Here are some images that have already been magnified and compressed in the Fractal Image Format (FIF).

(7ick on the magnification size to view

•of 3 06/20/96 08 52

:emonstration of Fractal Compression Technology http //xenon Proxima corn 8080/usgs/info/demo 1

4x 8x 16x

spcmsored by MCI

06/20/96 08 52

USGS LIBRARY RESTON

HI 1 11 Mill HI l HI H 3 1818 00242969 2