`of U.S. Pat. No. 7,477,284
`IPR2013‐00219
`EXHIBIT
`Sony‐
`
`
`
`THE PHOTOGRAMMETRIC JOURNAL OF FINLAND
`
`VOL 12
`
`No. 2
`
`1991
`
`Published by
`The Finnish Society of Photogrammetry and Remote Sensing
`and
`Institute of Photogrammetry and Remote Sensing
`Helsinki University of Technology
`
`Editor
`Prof.Dr. Einari Kilpela
`
`Assistant Editor
`Ms. Anita Laiho
`
`Contents
`
`Preface
`
`,.
`
`Photogrammetric News
`
`Articles:
`
`AZIZI, A.:
`
`Implementation of A Digital Stereo-Photogrammetric
`System Based on the Use of Pre-Rectified Images
`
`HAKKARAINEN, J.:
`
`Pseudo-Resolution
`
`HEIKKILA, J.:
`
`Use of Linear Features in Digital Photogrammetry
`
`HELAVA , U. :
`
`Prospects in Digital Photogrammetry
`
`HELAVA, U.:
`
`State of the Art in Digital Photogrammetric Work(cid:173)
`stations
`
`3
`
`4
`
`7
`
`20
`
`40
`
`57
`
`65
`
`( continued on the next page )
`
`The Journal will be issued whenever the number of contributions warrants publication.
`Contributions will be in English or German. Those wishing to subscribe this Journal are
`asked to contact its publisher.
`
`All correspondence shou ld be addressed to :
`Helsinki University of Technology, Institute of Photogrammetry and Remote Sensing,
`Otakaari 1, 02150 Espoo , FINLAND.
`
`Te lex : 125 161 htkk sf
`
`Telefax :
`
`Int. + 358-0-465 077
`
`• "
`
`1
`
`
`
`Contents ( cont. )
`
`HCZ>GHOLEN, A.,
`JAAKKOLA, J.:
`
`The Influence of Temperature Variations on the
`Photo Coordinates in An Analytical Plotter
`
`KUITTINEN, R.,
`YU, X., SOTKAS, P.:
`
`Estimation of Interpretation Error in Remote Sensing
`
`,
`
`I
`
`77
`
`86
`
`2
`
`
`
`STATE OF THE ART IN DIGITAL PHOTOGRAMMETRIC WORKSTATIONS
`by
`U. V. Helava
`West Palm Beach, Florida, U.S.A.
`
`ABSTRACT
`
`5
`
`,
`
`s
`
`e
`
`d
`
`reached a remarkably
`Digital photogrammetry workstations have
`mature technical status. All basic problems have been solved and
`brought
`to practice. The objective of this paper is to list the
`most important technical achievements of recent years, so as
`to
`define the state of the art as of today. The list results from an
`examination of the developments in which
`the author has been
`personally
`involved.
`It contains
`twenty five state-of-the-art
`items. At the end of the paper, reasons are given for the expec(cid:173)
`tation
`that
`the photogrammetric workstation will disappear as a
`specially designed entity.
`
`INTRODUCTION
`
`tre(cid:173)
`has grown
`During the past half a century, photogrammetry
`mendously as an art and a science, and become a dominant mapping
`tool all over the world. In the field of photogrammetric
`instru(cid:173)
`ments, we have observed the rise -- and decline --
`of "optical
`train" stereo plotters, the birth of analytical triangulation and
`block adjustment, and the ascent of analytical plotters to their
`current position as preferred photogrammetric instruments.
`
`that of Digital
`We are now witnessing the advent of a new era,
`Photogrammetry.
`Its
`roots are in computer technology, computer
`imaging and analytical photogrammetry. I've heard it said that
`a
`digital photogrammetry workstation
`is a new type of analytical
`plotter. In a limited sense that may be so, but
`the concept of
`digital photogrammetry has much greater potential. It brings a(cid:173)
`bout a bigger revolution than any instrument or methodology deve(cid:173)
`lopment
`in photogrammetry up to now. When it is fully developed,
`its effects will be felt also in the associated disciplines
`such
`as Remote Sensing and Geographic
`Information Systems <GIS>.
`Photogrammetry itself will be dramatically different.
`
`technical status of
`the present
`review
`I'll
`In this paper,
`digital photogrammetry.
`I'll do
`it from a somewhat restricted
`personal perspective. I'll focus primarily on workstations
`and
`systems
`because
`I
`have been fortunate to have been able to
`participate in major developments in
`those aspects of digital
`photogrammetry. These developments include systems for the U.S.A.
`Government and for the commercial market place, thus bracketting
`the entire cost and performance spectrum of workstations. I
`believe they represent the state of the art in systems developed
`for mapping purposes.
`Important developments have taken place
`in industrial digital photogrammetry, digital sensors, etc., but
`l
`am not
`in
`a position to deal with them in detail. After the
`review of the state of the art, I'll discuss briefly my under(cid:173)
`standing of
`the trends in workstation developments -- and why I
`
`65
`
`•
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`think the "photogrammetric workstation" will become nothing more
`than sofware running on a standard workstation.
`
`STATE OF THE ART IN DIGITAL PHOTOGRAMMETRY
`
`First, a little history. The concept of a digital <softcopy> pho(cid:173)
`togrammetric workstation has been around
`for approximately
`ten
`years. Publications by Tapani Sarjakoski <1981) and James Case
`(1982> mark the beginning of the digital photogrammetry era, al(cid:173)
`though several other people have worked earlier on related sub(cid:173)
`jects. Dr. Case described a digital workstation in his paper en(cid:173)
`titled
`"The Digital Stereo Comparator/Compiler <DSCC)"
`<Case,
`1982>. The description led to the development of the first digi(cid:173)
`tal photogrammetry workstation. Detailed specifications for the
`workstation were defined when the U.S.A. Government placed a con(cid:173)
`tract
`for
`the development of the DSCC. The contract was won by
`the General Dynamics/Helava Associates team. Over one dozen DSCCs
`were delivered
`in 1985. Several variants have been developed
`since then. Hundreds of such workstations are under contract,
`scores have been built and are in operation.
`
`To my knowledge, the DSCC was the first fully operational photo(cid:173)
`grammetric "softcopy workstation". Its development project was
`very ambitious;
`the
`resulting instrument had to meet a compre(cid:173)
`hensive set of capability and performance requirements. Even
`though it was the first one, it is still a state-of-the-art work(cid:173)
`station, except <perhaps> for its display system. Therefore, I'll
`describe the DSCC in some detail to establish a reference against
`which the later developments can be compared. I'll comment more
`briefly on the General Dynamics in-house experimental system, the
`Digital Image Workstation Suite <DIWS>, the DMA Workstation,
`the
`HAI-500,
`the HAI-750, and some significant ancillary devices to
`give a fuller picture of the state of the art. I'll not deacribe
`many
`technical details;
`rather, I'll identify state-of-the-art
`features in each, with the objective of developing a list of such
`features "for the record".
`
`Digital Stereo Comparator/Compiler <DSCC>
`
`a group of DSCC workstations. The workstations
`shows
`1
`Figure
`have binocular viewing systems and the general appearance of
`a
`large analytical plotter. The computer system driving the work(cid:173)
`stations is not shown. It consists of several six feet high cab(cid:173)
`inets of computer equipment. Each workstation has 56 special com(cid:173)
`puter boards, designed and built by General Dynamics,
`and
`a
`Floating Point Systems 5310 array processor. Two workstations
`share a VAX 11/780 computer. There is also a
`substantial
`amount
`of other computer equipment
`shared by all the workstations in
`the cluster shown in Figure 1.
`
`c
`1
`~
`
`F
`
`p
`
`F
`
`N
`r
`d
`
`G
`
`p
`1
`s
`g
`n'
`
`66
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`n
`
`v
`s
`j
`
`s
`
`n
`
`1
`t
`
`e
`e
`0
`e
`t
`h
`
`s
`a
`
`The Digital Stereo Comparator/Compiler <DSCC> was designed and
`built to strict specifications. It is evident that, in the minds
`of
`the specifiers, the DSCC was visualized as an analytical ste(cid:173)
`reoplotter for digital images. The specifications were demanding.
`The
`new photogrammetric workstation was expected to exceed the
`capabilities of the most advanced analytical plotters. When
`the
`DSCC was built,
`it met the specifications and verified the ex(cid:173)
`pectations placed on digital photogrammetry.
`
`The main specifications are summarized below in
`
`''bullet" form.
`
`Orientations
`
`Manual, semi-automatic and automatic measurement of fiducials,
`reseaus,
`and control points to better than 0.3 pixel precision;
`least squares interior orientation; bundle adjustment exterior o(cid:173)
`rientation; computation time less than 2 seconds.
`
`Data collection
`
`and semi-automatic measurement of points to subpixel pre(cid:173)
`Manual
`cision; automatic driving to known points; manual collection of
`contours and
`features; feature collection with on-line correla(cid:173)
`tion; automatic DTM collection with repeatability better than 0.3
`pixels
`in static mode and 1.41 pixels in dynamic mode; operator
`designated matrix structure, starting point, etc.
`
`Data editing
`
`Manual and semi-automatic positioning; any
`value can be edited; graphics superimposed
`
`previously
`in stereo.
`
`collected
`
`Photometric enhancement
`
`least 256 operator selectable, 7x7 pixel, axially symmetric,
`At
`individual filters.; tonal look-up tables with
`reloadable break(cid:173)
`points.
`
`Rotation and zooming
`
`changes in x and y parallaxes permitted; continuos, real-time
`No
`rotation with resampling over a range of 360 degrees; instant 90
`degree
`jumps at operator's request; continuous real-time zooming
`with resampling from 40x minification to 5x magnification.
`
`Graphics superimposition
`
`Points and lines superimposed on images in stereo as data is col(cid:173)
`lected; graphics resampling locked to image resampling; graphics
`superimposed on overview display at operator's request; DTM
`fi(cid:173)
`gure of merit shadings superimposed on overview, updated conti(cid:173)
`nuosly.
`
`67
`
`..
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`Image display
`
`Stereo display 512x512 pixels; overview display 1024x1024 rixels
`in color.
`
`System speed
`
`to 200 pixels per second while maintaining
`roam up
`speed
`Slow
`full stereo; up to 1000 pixels per
`second without maintaining
`stereo, correspondingly faster on minified images in terms of 1:1
`pixels; stereo jump in less
`than
`two
`seconds; DTM collection
`speed 200 points per second minimum, 500 points per second goal.
`
`System capacity
`
`image size 400 million pixels; at least two full images
`Maximum
`and minified versions stored simultaneously; additional storage
`for 50 1Kx1K image patches.
`
`Embedded in the specifications are numerous items that are parti(cid:173)
`cularly indicative of the state of the art. Some others evolved
`during
`the development. Here are the most significant ones, with
`comments.
`
`real-time
`(1) Subpixel pointing. The image processor performs
`resampling on stereo images to 1/128 of a pixel. Manual pointing
`is absolutely smooth in three dimensions. No "pixel jump" can be
`observed.
`In automatic correlation, the image patches are posi(cid:173)
`tioned with the same precision.
`
`(2) Full model roaming in stereo. The operator can move around
`in 3-D over
`the entire stereo model at plotting speed without
`losing stereo. The stereo model may include up to BOO Megapixels.
`A
`jump
`to anywhere in the model or photo can be accomplished in
`less than two seconds.
`
`<3> Comprehensive real-time image processing. The displayed ima(cid:173)
`ges are rectified or "dewarped", resampled, subjected to tonal
`adjustments and filtered sixty times per
`second. These actions
`can be controlled by the operator or by the software. In automa(cid:173)
`tic correlation, the image patches are rectified and
`filtered
`hundreds of times per second to achieve optimum results. The pa(cid:173)
`rameters for automation are commanded by
`the software and are
`different from those used for the visual displays.
`
`"Perfect" graphics superimposition. Cartographic graphics
`(4)
`are superimposed in stereo on the stereo display, and in mono
`on
`the overview display. Both are done in color. The graphics can be
`selectively faded in and out. Several state-of-the-art processes
`are included:
`(a) Graphics information in ground coordinates is projected back
`to images in real time.
`
`n
`
`r
`F
`s
`n
`1
`5
`
`p
`0
`t
`n
`
`Tt
`b!
`te
`
`Fi
`<E
`an
`t i
`ri
`an
`qu
`
`68
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`(bl The projection includes zooming and is accurate
`level.
`<c> Graphics track image motions to subpixel level.
`(d) Special processing removes line jaggedness.
`These processes assure
`that
`the
`superimposed graphics are in
`nearly perfect coincidence with image details at all times.
`
`subpixel
`
`to
`
`(5) Advanced automation. The DSCC has a multifunction correlator
`working
`in point-by-point mode, continually as
`the operator
`tracks features, or as a DTM collector. In all modes, the corre(cid:173)
`lator uses highly adaptive "strategies" and "tactics". Parameters
`involved in these are selectable by the operator, but are often
`set
`and modified on line by automation software. The DTM collec(cid:173)
`tion mode is particularly powerful. It is based on the concept of
`"Hierarchial Relaxation Correlation". DTM points are measured at
`hundreds of points per second, with high reliability and accura(cid:173)
`cy. Progress of DTM generation is shown on the overview display.
`Different colors are used to indicate the "figure of merit" on
`automatic measurements in different areas.
`
`DTM editing tools. Automatic DTM generation methods are not
`(6)
`yet capable of producing one hundred percent perfect results. So(cid:173)
`me manual editing will probably be necessary for a long time to
`come. The DSCC is equipped with a set of tools for making the ed(cid:173)
`iting as efficient as possible.
`
`triangulation. This was not a part of the o(cid:173)
`Semiautomatic
`(7)
`riginal DSCC, but was installed later as
`a
`retrofit.
`It uses
`Forstner's
`interest operator to select points, multi-image least
`squares correlation and hierarchical
`("image pyramid")
`tech(cid:173)
`niques. Several point candidates are measured at transfer point
`locations and tested for consistency before the
`final point
`is
`selected.
`
`Comprehensive math models. The great computer power of DSCC
`(8)
`permits realization of comprehensive math models. It is capable
`of handling a variety of geodetic projections and employs an ex(cid:173)
`tremely flexible method
`for
`representing
`image geometries
`in
`real-time computations.
`
`The DSCC has several other interesting features
`be state of the art but, as they are only more
`technologically, I'll ignore them here.
`
`which may well
`less
`trivial
`
`or
`
`General Dynamics Experimental Workstation <EW>
`
`the General Dynamics Experimental Workstation
`shows
`Figure 2
`<EW>. Its appearance is dramatically different from the binocular
`analytical plotters. The appearance of the EW signals the transi(cid:173)
`tion from a "stereoplotter" to a "workstation". All photogrammet(cid:173)
`ric softcopy workstations designed since have the general appear(cid:173)
`ance of the EW. The workstation is driven by DSCC
`computer e(cid:173)
`quipment, except for several "ad hoc" experimental additions.
`
`69
`
`-
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`General Dynamics designed and built the experimental workstation
`in 1985-86 as a platform for the investigation of new
`technolo(cid:173)
`gies and concepts. The workstation is probably the most advanced
`softcopy photogrammetric workstation in existence.
`
`system con(cid:173)
`The EW evolved from the DSCC. It fits into the DSCC
`figuration as a replacement for one of the binocular workstations
`of the DSCC. The EW actually shares a computer
`system with one
`binocular station located in the General Dynamics image process(cid:173)
`ing laboratory, where further research is taking place.
`
`The EW's contributions to the state of the art are listed below,
`with comments.
`
`The EW uses time-shared
`system.
`Free-view stereoviewing
`<9>
`the stereo
`images are alter(cid:173)
`stereo viewing. In this system,
`natingly displayed on a 512x512 pixel black and white monitor at
`the rate of 60 image pairs per second. The stereo separation
`is
`realized by a liquid crystal screen that changes polarity in syn(cid:173)
`chrony with the display. The screen is placed in front of the mo(cid:173)
`nitor
`and
`the observer wears polarized eye glasses. The liquid
`crystal screen was supplied by Tektronix, the drive and display
`electronics were designed
`and built by General Dynamics. This
`time-shared stereo display has been in use since mid 1986 without
`any problems. Similar displays are now in common use, e.g. in
`several GD/HAI workstation designs.
`
`(10> Generation of digital orthophotographs. Although DSCC has
`enough computational facility to produce digital orthophotographs
`very efficiently, it was not required to do so. Since the EW has
`the same facitilty, it was quite easy to program it for efficient
`orthophoto production. Orthophotos can be produced from vertical,
`convergent and oblique photographs, as well as from SPOT images.
`
`<11> Generation of perspective views. Impressive "very high ob-
`lique" perspective views can be routinely produced using the EW.
`Such views of "cityscapes" with many high rise buildings are par(cid:173)
`ticularly interesting. Ordinary vertical aerial photographs serve
`as inputs. When tall buildings are imaged somewhat off the nadir,
`the sides of the buildings are visible on the photographs. The EW
`perspective view generator transforms the images of the sides
`to
`fit
`the corresponding walls in the perspective views. The result
`is strikingly realistic. The realism can be enhanced
`further by
`"planting
`trees''. The need for that artifice is caused by diffi-
`culties in distinguishing the sides of trees on the input photo-
`graphs. The
`EW perspective view generator can "grow" trees for
`the operator to place on the scene as needed. The
`software a~-
`commodates several
`input photographs in which the city is seen
`from different sides. Series of perspective views can be
`then
`produced. Each view
`takes a few seconds to generate on the EW.
`They can be stored on a disk or on video tape and played back as
`movies,
`to
`(for example> simulate helicopter flights around the
`city.
`
`1 I
`
`a
`c
`n
`c
`t
`
`p
`t
`v
`c
`1
`t
`o
`b
`s
`a1
`
`J,
`a
`
`70
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`<12> Generation of stereomates. The generation of orthophotos
`and perspective views leads naturally to the essentially identi(cid:173)
`cal process of generation of stereomates. Both natural and arti(cid:173)
`ficial
`stereomates can be made, using various kinds of aerial
`photographs or SPOT images as inputs.
`
`(13) Preprocessing of images for lower cost workstations. This
`approach was
`studied during the design of the DSCC. For photo(cid:173)
`grammetric purposes, most image processing need not be
`in
`real
`time. Thus, one powerful workstation could preprocess imagery for
`compilation workstations dedicated to
`interactive data extrac(cid:173)
`tion. Several
`levels of preprocessing are feasible, from decom(cid:173)
`pression and reformatting to orthophoto/stereomate p~irs. This
`approach evolved
`to
`a
`stepwise processing used for example in
`HAI-500 commercial workstation.
`
`<14> Software image compression/decompression. The EW was pro(cid:173)
`grammed
`to perform image compression/decompression using various
`algorithms. Sophisticated algorithms permit compression of
`typi(cid:173)
`cal aerial photographs by a factor of four, without virtually any
`visual degradation upon decompression.
`
`to
`<15) Advanced operator controls. The EW was used extensively
`investigate operator controls. One of the innovations incorporat(cid:173)
`ed into the EW is a touch panel display/control. The panel
`is
`programmable. It can be used to display values of coordinates, o(cid:173)
`rientation elements, etc. simultaneously with a matrix of "but(cid:173)
`tons"
`that control the operation of the system. Each button in(cid:173)
`cludes text information about the action it controls. By touching
`a button, the operator causes the action to happen. Some buttons
`evoke new configurations of the panel, somewhat like bringing
`in
`a new menu on a CTR monitor. The touch panel has been found to be
`quite practical for some operating modes. For some operations, it
`needs
`to be supplemented with physical buttons at the coordinate
`control device. This enables the operators to keep their hands on
`the device instead of touching the panel.
`
`a movable cursor. A fixed cursor
`Subpixel pointing with
`<16>
`<movable image) mode places high demands on the image processor,
`particularly when fractional pixel pointing is required -- which
`the DSCC/EW image processor permits to 1/128 of a pixel. The mo(cid:173)
`vable cursor mode reduces image processing requirements dramati(cid:173)
`cally. The images need to be processed only once for potentially
`lengthy period of measurements. The drawbacks are the necessity
`to do the data extraction ''window-by-window", and the difficulty
`of
`implementing
`subpixel measurements. In the EW this is solved
`by providing a set of convolved cursors. The system automatically
`selects
`the cursor that best represents the subpixel position at
`any moment.
`
`that
`Just like the DSCC, the EW has several interesting features
`are not particularly state-of-the-art. I'll ignore them here.
`
`s
`
`t
`
`r
`
`n
`n
`
`s
`e
`
`---~-----------------------
`
`71
`
`..
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`The DMA Workstation
`
`The Defence Mapping Agency <DMA> workstation was designed as a
`component in a major upgrading effort started by the DMA
`in
`the
`early 1980s. This effort entails the implementation of a compre(cid:173)
`hensive system design in which different parts are
`tightly cou(cid:173)
`pled. Consequently,
`stand-alone operation of these workstations
`is not possible; they depend critically on other parts of
`the
`system for their inputs and outputs. This might sound like a step
`backwards. However, separation of functions is often desirable in
`large organizations. Such separation may well become a trend.
`
`DMA workststions are grouped into clusters of several worksta(cid:173)
`tions. Each cluster has substantial "shared
`resources"
`that
`is.
`computer power that serves all workstations in the cluster.
`Each workstation has a MicroVAX 3 as the host computer, ten 68020
`microprocessors
`and
`two array processors. Most image processing
`functions are moved to micros and array processors. Only
`a
`few
`custom designed boards are needed.
`
`Several state-of-the-art capabilities are present
`workstation cluster. They are listed below.
`
`in the DMA
`
`<17> Extremely large and fast mass storage of images. The stor(cid:173)
`age module, called the Cluster Image Subsystem <CIS>, consists of
`44 sychronized Winchester disks and the associated electronics.
`The
`system capacity is over 11 Gigabytes. The disks are operated
`in parallel, with each disk storing just one bit of the recording
`"word''. The word includes error detection bits. They are monito(cid:173)
`red continuously. If one disk crashes, the word can still be
`re(cid:173)
`constructed.
`In fact, the corrupt disk can be replaced, without
`shutting the system down. The information on that disk
`is
`then
`automatically
`reconstituted while the system runs. The parallel
`operation also gives the system high transfer rates, 360 Mega(cid:173)
`bits peak and 274 Megabits on average per second. The CIS is ca(cid:173)
`pable of serving image data to eight workstations.
`
`( 18) On-line image decompression. Each DMA workstation
`includes
`on-line
`image decomperession hardware. Use of compressed images
`increases the storage capacity of
`the CIS
`and
`improves
`the
`transfer rate to the workstations.
`
`feature extraction. The feature ex(cid:173)
`for automated
`Tools
`(19>
`traction tools are designed to increase production in collecting
`digital
`feature data. The array processors are used to perform
`automatic feature delineation in three dimensions using a
`combi(cid:173)
`nation of machine vision
`tools. These include region growing,
`pattern following and correlation. The automatic tools can work
`simultaneously with the operator or independently.
`
`Knowledge base feature attribution. Attributes must be as(cid:173)
`<20>
`signed to features, in order to make the data meaningful in a di(cid:173)
`gital data bank, such as a GIS. The process is very labor inten(cid:173)
`sive. DMA workstations have knowledge bases and associated soft-
`
`p
`"
`1
`
`t
`i
`E
`t
`
`1
`t
`Cl
`a
`I
`a
`
`c
`f
`g
`t
`a
`g
`
`h.
`t
`o·
`t
`Si
`
`c
`P•
`
`72
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`ware to assist the operator. The knowledge bases have information
`about features, their classification, relations between features,
`possible co-occurence, etc. The
`system
`is frequently able to
`"guess" what the feature might be and present the operator with
`lists of alternatives and as choices to refine the guess.
`
`(21> High resolution, high contrast stereo display. DMA worksta(cid:173)
`tions have improved time-share stereo displays. The display area
`is 1024x1024 pixels and the polarizing shutter is custom made to
`ensure high contrast images and an improved extinction ratio
`for
`better stereo separation.
`
`features present in DMA
`less significant
`number of
`a
`Again,
`workstations are not listed in this review of the state of
`the
`art.
`
`Digital Image Workstation Suite <DIWS>
`
`Image Workstation Suite was designed and built for
`The Digital
`the U.S. Navy. It is a complete system, from input digitizers
`to
`output image printers. In appearance and in most of its technical
`aspects, the DIWS workstation is very similar to the DMA version.
`It does, however, have its own contributions to the state of the
`art.
`
`is capable of ac(cid:173)
`<22> Universal input unit. The DIWS system
`cepting
`images on a wide variety of input media and in different
`formats. In addition to having many input devices, it has a pro(cid:173)
`grammable decompressor/reformatter/compressor. The purpose of
`this unit is to accept input images compressed using virtually
`any
`known algorithm, decompress them, reformat and compress a(cid:173)
`gain, using an algorithm designed for internal use in the system.
`
`(23) Three terabyte image storage. The DIWS image storage system
`has a robotic device capable of fetching any one of 600 tape car(cid:173)
`tridges in a few seconds. Each cartridge contains five gigabytes
`of data. The selected cartridge
`is loaded into a reader that
`transfers the data to a "Mini CIS" disk stack. The disk
`stack
`serves several workstations.
`
`(24> High speed, high resolution image digitizer. One of theca(cid:173)
`pabilities of the DIWS is to accept film images. For
`that
`task,
`it has an image digitizer as a part of the system. The digitizer
`accepts rolls of film up to nine inches wide or
`individual
`film
`sheet up
`to 9x9 inches in size. The pixel size is 12.5x12.5 mi-
`. crometres and the digitizing speed 2 Megapixels per second.
`
`The workstations reviewed are all "full capability" workstations
`from
`the photogrammetric point of view. They can do everything
`top-of-the-line analytical plotters can do, and more. They all
`meet
`the most demanding specifications of the DSCC: full model
`roaming at high speed, on-line, real-time image filtering and de(cid:173)
`warping, and subpixel pointing. Also, they are all expensive.
`
`73
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`Commercial Workstations
`
`the HAI-500 is the first commercially offered
`knowledge,
`To my
`digital photogrammetry workstation. It has no custom made hard(cid:173)
`ware, only commercially available computer equipment. It runs on
`a 386-class PC, and is capable of producing DTMs <Digital Terrain
`Models>, orthophotographs and perspective views automatically. It
`has DTM editing tools, and permits manual point measurement, con(cid:173)
`touring, profiling and feature collection. The HAI-500 achieves
`this fairly substantial performance by performing
`an off-line
`preprocessing, including epipolar rectification. The HAI-750 is a
`higher performance workstation in this same class. HAI-500
`and
`HAI-750 are vary similar in appearence; Figure 3 illustrates how
`they look, when their computers are under the work
`surface and
`hidden from the view.
`
`These workstations make the following contribution to the state
`of the art:
`
`<25> Full photogrammetric capability using standard hardware. In
`recent years,
`this workstation class has captured the attention
`of the photogrammetric community. These workstations are charac(cid:173)
`terized by
`some compromises in performance to gain substantial
`cost reduction. The primary compromises are in image processing.
`Cost
`reduction
`is achieved by the use of commercially available
`computer equipment. Several cost/performance compromises
`and
`combinations have appeared
`in
`the market place, and more are
`likely. None have yet achieved state-of-the-art performance, but
`they can, and will.
`
`technical
`tremendous
`photogrammetry has experienced
`Digital
`progress in its half a dozen years of existence. Better under(cid:173)
`standing of technical problems involved in realizing workstations
`for digital photogrammetry, as illustrated by the above
`list of
`state-of-the-art capabilities,
`is
`a part of that progress. An(cid:173)
`other part is the "free ride" the field has had
`from computer
`technology. Capabilities
`that,
`a few years ago, took expensive
`custom equipment to realize can now be accomplished with "store(cid:173)
`bought", modestly priced hardware.
`
`State-of-the-art workstations for digital photogrammetry are low
`cost, highly automated and capable of doing everything
`the
`tra(cid:173)
`ditional photogrammetric
`instruments can do, and more. Manufac(cid:173)
`turers of photogrammetric instruments and practising photogram(cid:173)
`metrists ignore this at their peril.
`
`C(
`
`Ir
`a
`ni
`p<
`1~
`
`Ir
`ol
`pr
`pr
`ph
`ir
`th
`
`RE
`
`Ca
`Ir
`
`Sa
`St
`
`74
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`· - · · - - - - - - - - - - - - -
`
`GoTTERDAMMERUNG OF PHOTOGRAMMETRIC INSTRUMENTS
`
`out of my 45 year career in photogrammetry, I have spent over 35
`in developing photogrammetric instruments -- the last ten, trying
`to cast
`them
`into the form of digital workstations. It is with
`some nostalgic pain that I see the coming disappearance of pho(cid:173)
`togrammetric
`instruments. This includes digital workstations as
`entities specially designed and manufactured for photogrammetry.
`As
`far
`a
`I am concerned, the writing is on the wall. Photogram(cid:173)
`metric instruments
`including my
`favorite,
`the analytical
`plotter
`will be replaced by digital workstations. The photo(cid:173)
`grammetric "instrument" will soon be nothing more than a standard
`issue workstation and a box of shrink-wrapped software.
`toward preprocessing.
`
`CONCLUDING WORDS
`
`terms of technology, diQital image workstations have reached
`In
`a remarkable level of maturity. Solutions to all critical
`tech(cid:173)
`nical challenges have been developed and brought to practice. Ca(cid:173)
`pabilities of the workstations match -- or exeed -- those of ana(cid:173)
`lytical plotters.
`
`terms of commercial use, several challenges remain. The rev(cid:173)
`In
`olution is profound; it requires drastic rethinking of methods,
`procedures and practices. Digital photogrammetry ushers in new
`products, new ways of using map information and new approaches to
`photogrammetric production. The change will be particularly try(cid:173)
`ing to makers of instruments: they'll have find a
`new
`role
`for
`themselves.
`
`REFERENCES
`
`Case, J. 8., 1982. The Digital Stereo Comparator/Compiler <DSCC)
`International Archives of Photogrammetry, 24-II:23-29.
`
`a Conpletely Digital
`1981. Concept of
`Tapani,
`Sarjakoski,
`Stereoplotter, The Photogrammetri Journal of Finland, 2:95-100.
`
`75
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`
`
`Figure 1.
`A Cluster of DSCC Wor1<stations.
`
`------
`
`Figure 2.
`The General Dynamics
`Experimental Wor1<station.
`
`Figure 3.
`The HAI-500 or ltle HAI-750 Wor1<station.
`
`76
`
`}
`
`l
`1:
`to
`t
`
`1
`
`'I
`1
`I=
`c
`d
`e
`d
`a
`v
`n
`I<
`t
`s
`
`2
`
`T
`m
`d
`r
`r
`t
`d
`t
`t
`T
`r ·
`
`T
`
`The Photogrammetric Journal of Finland, Vol. 12, No. 2, 1991
`
`