Digital Camera Patent Abstract
A high resolution, high speed digital camera that produces a high
resolution image representative of an image scene viewed by the
camera. The digital camera comprises an input lens system and an
iris that couple light from the image scene into the camera. A first
imager generates a sequence of high resolution, low speed digital
frames of the image scene. A second imager generates a sequence
of low resolution, high speed digital frames of the image scene.
Image processing electronics process the digital outputs of the
respective imagers to register the high speed and low speed digital
frames to each other, generate a vector field that describes the
motion of pixels from one frame to the next in the sequence of low
resolution, high speed frames, and interpolate pixels of the next
frame in the sequence of high resolution, low speed frames using
the vector field to translate them in the same manner as the pixels
changes in the low resolution, high speed frames to produce high
resolution digital frames representative of the images in a scene
containing objects moving at high speed. Digital Camera Patent Claims
What is claimed is:
1. A digital camera for producing a high resolution image representative
of an image scene viewed by the camera, comprising:
a first imager for generating a sequence of high resolution, low
speed digital frames of the image scene;
a second imager for generating a sequence of lower resolution,
high speed digital frames of the image scene; and
image processing electronics coupled to the respective imagers
for registering the high speed and low speed digital images to each
other, for generating a vector field that describes the motion of
pixels from one frame to the next in the sequence of low resolution,
high speed frames, and for interpolating pixels of the next frame
in the sequence of high resolution, low speed frames using the vector
field to move them in substantially the same manner as the pixels
in the lower resolution, high speed frames to produce high resolution
digital images representative of the image scene.
2. The camera recited in claim 1 further comprising an input lens
system and an iris that couple light from the image scene into the
camera.
3. The camera recited in claim 1 wherein the vector field of pixel
motions between high speed frames is estimated using a decomposition
technique.
4. A digital camera for producing a high resolution image representative
of an image scene viewed by the camera, comprising:
a first imager for processing high resolution image frames recorded
at a low rate to record detail in the image scene;
a second imager for processing lower resolution image frames recorded
at a high rate to record the movement of pixels in the image scene;
and
image processing electronics coupled to the respective imagers
for processing the high resolution and lower resolution image frames
to produce a sequence of high resolution picture frames at a high
rate, and wherein the high resolution image frames capture the fine
details of the scene and the lower resolution image frames capture
motion of objects in the scene.
5. The camera recited in claim 4 further comprising an input lens
system and an iris that couple light from the image scene into the
camera.
6. The camera recited in claim 4 wherein the image processing electronics
uses digital data from the high speed, low resolution image frames
to interpolate the locations of moving objects as they transition
between the high resolution image frames.
7. The camera recited in claim 6 wherein the image processing electronics
generates intermediate high resolution frames that are inserted
between the high resolution frames to provide a sequence of image
frames that have the high resolution detail derived from the first
imager as well as motion dynamics derived from the second imager.
8. A digital camera for producing a high resolution image representative
of an image scene viewed by the camera, comprising:
an input lens system and an iris that couples light from the image
scene into the camera;
a beam splitter that couples light from the image scene along first
and second light paths;
at least one shutter that interrupts the first and second light
paths;
a plurality of beam splitters that separate the image along three
light paths;
color filters respectively disposed in the three light paths separate
out color spectra contained in the image
a high resolution, low speed imager that produces image truth-frames
disposed along each of the three light paths;
a low resolution, high speed imager that produces motion key-frames
disposed along the second path; and
image processing electronics coupled to the imagers for processing
the image truth-frames and motion-frames to produce high resolution
digital image output data representative of an image scene containing
objects moving at high speed.
9. The camera recited in claim 8 wherein the image processing electronics
processes the high resolution and lower resolution image frames
to produce a sequence of high resolution picture frames at a high
rate, and wherein the high resolution image frames capture the fine
details of the scene and the lower resolution image frames capture
motion of objects in the scene.
Digital Camera Patent Description
BACKGROUND
The present invention relates generally to digital cameras, and
more specifically, to a high resolution, high speed digital camera
for cinema and multimedia applications.
A digital camera is desirable for cinematography instead of film
due to its compatibility with the growing trends in the use of digital
techniques in the production of movies and other multimedia products.
In addition, a digital camera provides artistic flexibility and
economic benefits over the use of film cameras.
Existing digital cameras do not meet the speed and resolution requirements
of the movie industry. There are high speed digital cameras with
low to moderate resolution capability and there are high resolution
digital cameras with low to moderate speed capabilities. However,
there are none capable of both high resolution and high speed performance.
Digital camera complexity is proportional to the number of pixels
per frame, the number of bits per pixel, and the number of frames
per second provided by the camera.
Dicomed produces single shot high resolution digital cameras. These
products include the BigShot.TM. digital instant cameras, ScanBack.TM.
digital scanning cameras and StudioPro.TM. digital scanning camera
backs. Kodak produces high speed low resolution cameras including
the Megaplus.TM. cameras that have 15, 30 and 60 frames per second
frame rates with 1024 by 1024 pixel resolution.
None of these cameras meet the requirements for the movie industry.
Cinematography experts in the movie industry have defined the requirements
for a digital camera to have 4096 by 4096 pixel resolution, at least
10 bits digitization for each of the three primary colors, and a
speed of 60 frames per second.
A conventionally designed digital camera with these characteristics
would produce an output data rate in excess of 30 Gigabits per second.
This data rate is at the edge of technical feasibility from a speed
perspective. In addition, an expensive storage system with more
than 20 Terrabytes of capacity would be required to store a typical
90 minute movie. This places a conventionally designed digital camera
system at the limits of affordability and technical feasibility.
However, there are economic and artistic benefits that could accrue
from a system that can capture the live action in a scene in a digital
format with high detail and color and in nearly real time.
Accordingly, it is an objective of the present invention to provide
for a high resolution high speed digital camera for use in movie
and multimedia applications that is technically realizable and is
affordable. This objective will be achieved through a non conventionally
designed camera that reduces the data rate and reduces the required
storage capacity.
SUMMARY OF THE INVENTION
To accomplish the above and other objectives, the present invention
provides for a high resolution, high speed camera that produces
a high resolution digital color representation of a moving image
scene viewed by the camera. The camera captures several high resolution
frames per second and many more lower resolution frames per second.
The data from the frames are then used to produce a sequence of
high resolution picture frames at a high rate. The high resolution
frames capture the fine details of the scene while the low resolution
frames capture the motion of objects in the scene.
The camera comprises an input lens and shutter system that couples
light from the image scene into an electro-optic imaging system
located within the camera body. The electro-optic system is comprised
of two types of imaging sensors. A first imaging sensor generates
a digital output representing a high resolution sequence of frames
of the image scene at a relatively low frame rate. A second imaging
sensor generates a digital output representing a sequence of lower
resolution digital frames of the image scene at a high frame rate.
The digital outputs from the two types of imaging sensors are processed
by image processing electronics to produce a combined digital output
that is a high speed sequence of high resolution frames of the image
scene. The image processor uses the digital data from the high speed,
low resolution frames to provide an interpolation of the locations
of moving objects as they transition between the high resolution
frames. The interpolation algorithm is used to generate intermediate
high resolution frames, which are inserted between the captured
high resolution frames. This results in a sequence of color imagery
frames that have the full detail of the high resolution sensor as
well as the motion dynamics captured by the high speed sensor.
The data generated by the camera is the sum of the data from the
two sensors, which will in the order of 1/4 to 1/10 of the data
generated by a conventional digital camera with the same characteristics.
The actual improvement is a function of the ratio of the two frame
rates. This ratio is chosen to provide the desired artistic cinematic
effects.
The digital camera performs very high resolution color digital
imaging, at very high frame rates, on the order of 60 frames per
second or more. The resolution and speed performance of the high
resolution, high speed digital camera is comparable or exceeds the
performance of high quality film cameras.
The digital camera provides a nearly instantaneous viewable output
of captured images. The output may be examined to determine if changes
to the scene are needed and a retake can be made while the set is
still in place. When film is used the film processing time limits
the flexibility in making these kind of instantaneous changes.
The digital camera is designed to replace film cameras for making
movies and other multimedia applications. The digital camera provides
the means to make a revolutionary change in the way the movie industry
makes movies by eliminating film processing steps that are currently
employed.
BRIEF DESCRIPTION OF THE DRAWINGS
The various features and advantages of the present invention may
be more readily understood with reference to the following detailed
description taken in conjunction with the accompanying drawing,
which illustrates an exemplary high resolution, high speed digital
camera in accordance with the principles of the present invention.
DETAILED DESCRIPTION
Heretofore, high resolution digital cameras that operate at a high
frame rate have been difficult to build because of the high data
output rates that are required. A representative high resolution
image has on the order of 4 k.times.4 k pixels. Each pixel of a
color image has at least 10 bits each of red, green, and blue colors.
The size of the digital camera frame is (3.times.10 bits.times.16)
million pixels or 480 megabits per frame. Special effects such as
"slow motion" of high speed objects require cameras that
operate at 60 frames per second or higher. This requires an output
data rate of (60.times.480)=28.8 Gigabits per second for a high
resolution image.
Referring to the sole drawing figure, the present invention provides
for a digital camera 10 that embodies a technique that uses high
resolution images recorded at a low rate to record detail in the
images while using lower resolution images recorded at a high rate
to record the movement of the pixels in the images. This technique
provides a compromise between the effective data rates from the
two types of images. When reconstructed, the sequence of high rate
images produced from such an image set will have the detail of the
low rate, high resolution images and the frame rate of the high
rate, low resolution images.
A representative low resolution image is a 1 k by 1 k image gray
scale image. Only 12 bits are required for one pixel. Each frame
has 1.5 million bytes. The lower resolution gray scale images requires
a data rate from the digital camera that is 49 times smaller than
the large color image data rate.
Referring again to the drawing figure, it illustrates an exemplary
high resolution, high speed digital camera 10 in accordance with
the principles of the present invention. The exemplary high resolution,
high speed digital camera 10 comprises an input lens system 12 and
an iris 13 that couple light from an image scene 11 to a beam splitter
14. The beamsplitter 14 couples light from the image scene 11 along
first and second paths 20, 30 to a first shutter 21 and a second
high speed shutter 31, respectively.
Light transmitted by the first shutter 21 is coupled by a plurality
of beam splitters 22 that separate the image along three light paths.
Individual red, green and blue color filters 23 respectively disposed
in the three light paths separate out the red, green and blue color
spectra contained in the image. The respective images transmitted
by the red, green and blue color filters 23 are imaged onto high
resolution charge coupled device (CCD) imager 24 having drive electronics
that produce frames referenced as image truth-frames. In the alternative,
each of the high resolution CCD imagers 24 may process light from
the image having different shades of gray in lieu of separate color
components. The outputs of the CCD imagers 24 are input to image
processing electronics 27.
Each of the high resolution CCD imagers 24 produce image truth-frames.
The high resolution CCD imagers 24 have a resolution on the order
of 4 k by 4 k pixels which is comparable to film resolution. However,
the high resolution CCD imagers 24 only operate at about two frames
per second.
Light transmitted by the second high speed shutter 31 is reflected
off of a mirror 32 and imaged with a second lens system 33 onto
a high speed CCD imager 34 having drive electronics that produces
motion-frames. The output of the high speed CCD imager 34 are input
to the image processing electronics 27.
The image processing electronics 27 processes the image truth-frames
and motion-frames to produce high resolution digital image output
data 11a representative of the image scene 11. The processing performed
in the camera 10 is discussed in detail below.
The high resolution image are used as truth images that are updated
at the low resolution CCD sensor frame rate. Each pixel in the high
resolution data of a frame has a number that defines its location
and color or gray scale characteristics (typically 12 bits per pixel
for color). The number associated with each pixel changes when the
truth-image is updated. Between updates, frames of high speed data
are generated by the high speed CCD device The image processing
electronics 27 interpolates between two successive truth points
by using the high speed data to generate additional high resolution
image frames. Thus high speed vectors associated with each pixel
as data is processed added to the high resolution vectors associated
with each pixel as data is processed from frame to frame.
More specifically, an exemplary camera 10 may produce high resolution
frames at a rate of four per second and low resolution frames running
at 60 frames per second. The combined data rate of the exemplary
camera is 4.times.72+60.times.1.5=378 megabytes per second. This
data rate compares with 60.times.72=4310 megabytes per second from
a contemporary single high resolution imager operating at 60 frames
per second. This is shown in Table 1 below.
TABLE 1 Data rates for each resolution Size Rate Image MBytes Frames/Sec.
MB/Sec Mixed Resolution Images 4k .times. 4k 72 4 288 1k .times.
1k 1.5 60 90 Total 378 High Resolution Images 4k .times. 4k 72 60
4320
The data rate for the exemplary camera is less than 10% of the
data rate from a conventional digital camera.
The exemplary camera 10 produces acceptable frames at the full
rate that are constructed from a sequence of mixed rate, mixed resolution
frames. The construction of the high resolution, high rate frames
from the mixed resolution, mixed rate frames is a relatively straightforward
application of technology similar to image compression. The intermediate
frames between truth-frames are generated by predicting the pixel
values for each successive frame from the preceding frame. Forward
prediction is used for the two frames after a truth-frame, while
backward prediction is used for the two frames behind a truth-frame.
The frame that is midway between two truth-frames uses a combination
of forward prediction from one frame and backward prediction for
the next frame.
The intermediate images are encoded by taking the difference between
the predicted pixel values and the actual pixel values. The difference
image has much smaller values for the difference pixels than does
the original image. Consequently, the intermediate image can be
expressed as a much smaller digital packet.
The effective size of the intermediate images depends on the prediction
of the pixel values. The prediction is much more accurate when the
motion of the pixels is taken into account. An estimate of the motion
is constructed in the process of reconstructing the image sequence.
The motion is used to predict the value of the next image in the
sequence. The difference image is added back to the predicted image
to form the reconstructed image.
The prediction of the pixel motion depends on tracking an individual
pixel from one frame to the next. When the image contains moving
objects, the image of the moving object moves over the image area.
A vector field is generated over the area of the image indicating,
for each image, the motion of the pixels as they move to form the
next image in the sequence. A vector field that describes the motion
of the pixels from one image to the next in the sequence is generated
from the low resolution high rate images. The vector field is then
used to move the pixels of the high resolution image to the appropriate
locations in the next image in the sequence.
In the sequence of images shown in Table 1 there are 15 high speed
frames between each of the low speed frames. The high speed frames
are used to predict forward from one high resolution frame for seven
frames and backward from the next high resolution frame for seven
frames. The prediction for the motion carries a high resolution
pixel forward or backward by not more than seven frames.
For areas of the image where there is little or no motion, the
pixels move very little. For areas where there is rapid motion,
the pixels move some distance. The pixels with little motion are
easy to track through intermediate frames accurately. The rapid
motion pixels are more difficult to track. However, the error in
tracking is balanced against the blurring effect on the image that
occurs when the light collected for one pixel comes from several
areas due to motion. Moving objects are blurred in the high resolution
image because of the motion. The errors in prediction cause blurring
that is comparable to the blurring of any high resolution frame
due to the motion.
The processing of the two image sets must go through several steps:
(1) registration of the low resolution images to the high resolution
images; (2) generation of the motion vector field at each frame
time; and (3) interpolation of the pixels of the high resolution
image to locations determined by the pixel motion vector field.
The low resolution images must be registered to the high resolution
images so that the motion of pixels of the low resolution imager
can be related to motion of pixels of the high resolution image.
An interpolation function with a translation forms each pixel in
the next image frame.
The processing performed in the camera 10 uses a pixel motion vector
field. The motion of pixels from one frame to the next is determined
by a local registration of one frame compared to the next. With
regard to the concept of the vector field of motion, between two
frames in sequence, some of the pixels move a lot while others have
much less motion. For each pixel, a vector indicates where that
pixel has moved between the two frames. When the direction and magnitude
of the vector are known, the pixel value for the next high resolution
frame in the sequence is a translation of a pixel in the preceding
image.
At the sharp transitions between moving objects and the backgrounds,
the transitions are blurred because of the time required to form
an individual image of a moving object. However, the human eye does
not perceive detail in objects that are moving across the image.
Also, when the camera 10 is panned, the edges that come into view
are low resolution, corresponding to the low resolution of the high
speed camera 10. The result is not noticeable because of the insensitivity
of the eye during rapid motion.
The interpolation at the edges of moving objects may use information
from the high speed images to form pixels that are uncovered by
the moving object. Similarly, pixels that are covered by a moving
object are not used in the interpolation of new pixels in the next
image.
The treatment of the transition regions is different from the treatment
of the general areas of the image. The definition of the transition
regions is aided by finding the boundaries of objects. The process
of finding such boundaries is well known. The edges are discovered
by an edge finding processing. This scheme is basically a high pass
filter with processing to form lines establishing the edges.
When the edge of an object is established, the treatment of pixels
at that edge are adjusted depending on whether the pixel is part
of the moving object or part of the background behind the moving
object.
As discussed above, movement of the camera 10 results in an apparent
motion of the entire scene. This motion is determined by the lowest
resolution phase determination. The apparent motion is taken out
of the pixel translation process easily, leaving only the relative
motion of objects in the images.
Thus, a high resolution, high speed digital camera for use in film
and multimedia applications has been disclosed. It is to be understood
that the above-described embodiment is merely illustrative of some
of the many specific embodiments that represent applications of
the principles of the present invention. Clearly, numerous and other
arrangements can be readily devised by those skilled in the art
without departing from the scope of the invention.
|