Image analysis – Color image processing
Reexamination Certificate
2001-05-10
2004-07-27
Boudreau, Leo (Department: 2621)
Image analysis
Color image processing
C358S001900, C348S273000
Reexamination Certificate
active
06768815
ABSTRACT:
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The invention relates generally to the field of color sensors and more particularly to color sensors having neural networks with a plurality of hidden layers, or multi-layer neural networks, and further to a new neural network processor for sensing color in optical image data.
(2) Description of the Prior Art
Electronic neural networks have been developed to rapidly identify patterns in certain types of input data, or accurately to classify the input patterns into one of a plurality of predetermined classifications. For example, neural networks have been developed which can recognize and identify patterns, such as the identification of hand-written alphanumeric characters, in response to input data constituting the pattern of on and off picture elements, or “pixels”, representing the images of the characters to be identified. In such a neural network, the pixel pattern is represented by, for example, electrical signals coupled to a plurality of input terminals, which, in turn, are connected to a number of processing nodes, each of which is associated with one of the alphanumeric characters which the neural network can identify. The input signals from the input terminals are coupled to the processing nodes through certain weighting functions, and each processing node generates an output signal which represents a value that is a non-linear function of the pattern of weighted input signals applied thereto. Based on the values of the weighted pattern of input signals from the input terminals, if the input signals represent a character that can be identified by the neural network, the one of the processing nodes associated with that character will generate a positive output signal, and the others will not. On the other hand, if the input signals do not represent a character that can be identified by the neural network, none of the processing nodes will generate a positive output signal. Neural networks have been developed which can perform similar pattern recognition in a number of diverse areas.
The particular patterns that the neural network can identify depend on the weighting functions and the particular connections of the input terminals to the processing nodes. The weighting functions in, for example, the above-described character recognition neural network, essentially will represent the pixel patterns that define each particular character. Typically, each processing node will perform a summation operation in connection with values representing the weighted input signals provided thereto, to generate a sum that represents the likelihood that the character to be identified is the character associated with that processing node. The processing node then applies the non-linear function to that sum to generate a positive output signal if the sum is, for example, above a predetermined threshold value. Conventional non-linear functions which processing nodes may use in connection with the sum of weighted input signals is generally a step function, a threshold function, or a sigmoid, in all cases the output signal from the processing node will approach the same positive output signal asymptotically.
Before a neural network can be useful, the weighting functions for each of the respective input signals must be established. In some cases, the weighting functions can be established a priori. Normally, however, a neural network goes through a training phase, in which input signals representing a number of training patterns for the types of items to be classified, for example, the pixel patterns of the various hand-written characters in the character-recognition example, are applied to the input terminals, and the output signals from the processing nodes are tested. Based on the pattern of output signals from the processing nodes for each training example, the weighting functions are adjusted over a number of trials. After the neural network has been trained, during an operational phase it can generally accurately recognize patterns, with the degree of success based in part on the number of training patterns applied to the neural network during the training stage, and the degree of dissimilarity between patterns to be identified. Such a neural network can also typically identify patterns that are similar, but not necessarily identical, to the training patterns.
One of the problems with conventional neural network architectures as described above is that the training methodology, generally known as the “back-propagation” method, is often extremely slow in a number of important applications. In addition, under the back-propagation method, the neural network may result in erroneous results that may require restarting of training. Even after a neural network has been through a training phase, confidence that the best training has been accomplished may sometimes be poor. If a new classification is to be added to a trained neural network, the complete neural network must be retrained. In addition, the weighting functions generated during the training phase often cannot be interpreted in ways that readily provide understanding of what they particularly represent.
Edwin H. Land's Retinex theory of color vision is based upon “three color” experiments performed before 1959. A simple “mishap” showed that three colors were not always required to see accurate color. Land used a short and long record of brightness data (black and white transparencies) to produce color perceived by human eyes and not by photographic means. He demonstrated a perception of a full range of pastel colors using two very similar in color light sources such as yellow, at 579 nm and yellow orange, at 599 nm (“Experiments in Color Vision”, Edwin H. Land, Scientific American, Vol. 200 No. May 5, 1959). Land found that in some two record experiments all colors present were not perceived. Although Land demonstrated that two records provided color perceptions, he constructed his Retinex theory upon three records such as his long, medium and short records (An Alternative Technique for the Computation of the Designator in the Retinex Theory of Color Vision”, Edwin H. Land, Proceedings of the National Academy of Sciences, Vol. 83, 1986). The invention herein is related to human color perception discovered during Land's color vision experiments as reported in 1959.
The “Trichromatic” theory in human color vision has been accepted on and off since the time of Thomas Young in 1802 (A Vision in the Brain”, S. Zeki, Blackwell Scientific Publishing, 1993). Still and video electronic camera designs are correctly based upon the trichromatic theory but the current designs are highly subjective to color error reproduction due to changes in the ambient light color temperatures and color filtrations. The device in this invention senses color using a new “bichromatic” theory, which includes a mechanism that insures color constancy over a large range of ambient color temperatures. The use of two lightness records as used by Land in 1959 is one key to this invention.
The bichromatic theory is based upon an interpretation of a biological color process that occurs in the eyes and brain of humans and in some animals. The bichromatic theory is defined as a system that functions together under the following assumptions, accepted principles and rules of procedure, for which
FIGS. 4A and 4B
are provided for support:
(1) The system is a color sensing retina. There are at least two photo transducers in each pixel space in the retina, shown in
FIG. 4B
as TR(HI) and TR(LO).
(2) The two photo transducers sense the color of the light at each pixel's position in a scene of color focused on the retina. Each of the at least photo transducers contains a different spectral response and the wavelength difference between the peaks of a pair of these responses is called the waveband or the spectral bandwidth of the two photo transducers.
(3) The two photo transducers have overlapping spectral logarithmic responses where their slopes are opposing each other as indicated in FIG.
4
A.
(4) The photo
Boudreau Leo
Kasischke James M.
Nasser Jean-Paul A.
Oglo Michael F.
Sherali Ishrat
LandOfFree
Color sensor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Color sensor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Color sensor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3193304