Television – Monitoring – testing – or measuring
Reexamination Certificate
1999-11-08
2003-11-25
Miller, John (Department: 2614)
Television
Monitoring, testing, or measuring
C348S135000, C348S089000, C348S095000, C348S188000, C382S110000, C382S162000, C358S504000
Reexamination Certificate
active
06654048
ABSTRACT:
This invention relates to the calibration of imaging systems, particularly for systems used for capturing colour information for meat or carcass sections for subsequent analysis.
There have been over recent years attempts to use more objective systems analysing meat sections such as carcasses in an abattoir for the purpose of grading. Attempts have been made to capture by means of colour cameras colour information for automated analysis.
Patent specification No. WO-91/14180 describes in some detail the background of the necessity and known practices to evaluate a carcass in an abattoir e.g. for predicting yield. That specification discloses the capture of an image and colour analysis to discriminate fat, meat, bruised tissue, etc. The specification recommends uniform controlled lighting conditions.
Patent specification WO-92/00523 discloses a carcass grading system using video cameras to capture images. This specification recognises the need for calibration to a defined set of environmental characteristics but merely observes that new parameters are needed if conditions change radically. The specification proposes that it is possible to have a “dynamically self-calibrating” system upon start up, and a system which is “self-compensating in operation” but without describing how.
Patent specification WO-95/21375 discloses a system for assessing meat pieces in an enclosure where there is controlled illumination and a controlled temperature. The system refers to the provision of reflection and colour references to enable adjustment of the system for instability in the camera or in the light sources.
Patent specification No. EP0730146 discloses a system using controlled illumination in which an image of the background is captured and from this an image of a carcass overlying the backgrounds is subtracted. The specification refers to the system being sensitive to camera variation and to the need for controlled illumination and to the consistency of position between the two images. Reference is made to the position of colour pads placed in the field of vision to enable adjustment of the camera, the lamps and/or RGB values.
All of these patent specifications gloss over the provisions for calibration, possibly because it was not recognised how critical calibration might be or alternatively the systems have not been developed sufficiently for the calibration problems to have been seriously addressed and adequate provisions made.
In developing a system to enable reliable prediction of the yield (or other characteristics) of a carcass, we have been attempting to rely on the recognition that colours of the carcass surface (which vary depending on whether there is fat or meat or selvedge exposed) are some of the best predictors of a carcass yield. We have been testing a system which uses the average colour or RGB values from a number of selected sections or “patches” on the surface of a carcass to predict the amount of fat cover automatically and, we have been using empirically obtained data to derive statistical equations relating the measured colour or RGB values to predict yield. Preferably multiple patches at different areas of the carcass are imaged, the selected patches being those which have been determined to have correlations with the actual bone out yield, and the system measures the average colour on each of these patches and these data are combined into the yield prediction equation. Patches with higher correlations with the actual bone out yield can have higher coefficients in the equation than those with a lower correlation. Basically the whiter the colour the more fat that the system infers.
Colour is difficult to measure absolutely. The colour seen by the human eye or by a video camera varies depending on the amount of and the colour of the light reflected from the viewed surface. We have found that whilst the absolute colour is not critical or essential to the successful operation of the yield prediction system, the relativity of the colour is desirably consistent from one plant or abattoir to another. If this is not the case, the system when used in different plants would predict different yields for identical carcasses which, of course, defeats the purpose of attempting to have a computerised or automated grading or classifying system. It follows therefore th at calibration of an imaging system for use in automated analysis of products such as meat carcasses becomes extremely important to provide consistent results in different environments of use.
Ac cordially it is an object of the present invention to provide a method and apparatus for calibration of an imaging system which can ensure or at least promote consistency of operation in different environments .
It is a further preferred object to provide a method and system for imaging having effective calibration provisions for the system to promote consistent results under different operating conditions.
According to a firsts aspect of the present invention there is provided a method for calibration of an imaging system having a camera for capturing colour data for a viewed scene, the method including a primary calibration to enable compensation for variations in the camera operation or characteristics, the primary calibration including providing a number of standard colour specimens and presenting these to the camera under controlled illumination conditions, capturing image data for the standard colours including colour data and calculating a primary transform comprising compensation factors or transformation parameters for relating the actual measured colour data to known standard colour data for the particular standard colour specimens; the method further including a secondary calibration to enable compensation for local lighting conditions during capture and processing of colour data for captured images to be analysed, the secondary calibration process comprising presenting to the camera reference colour specimens exposed to the lighting conditions that the object to be imaged and analysed is being or is to be exposed, followed by capturing image data for the reference colour specimens and determining a secondary transform comprising compensation factors or transformation parameters for relating the measured colour data for the reference colour specimens to known colour data for those specimens.
According to a second aspect of the present invention there is also provided a method for imaging a target object by an imaging system and providing corrected image data for the target object, the method including the steps of: calibrating the imaging system according to the first aspect; capturing image data for the target object; and applying the secondary transform and the primary transform to the image data for the target object so as to yield more objective colour data for the target object.
In use where the target object comprises a carcass to be graded or classified according to colour characteristics, the method of imaging preferably comprises locating the carcass at an image station adjacent to which are provided the reference colour specimens so that the carcass and reference colour specimens are imaged simultaneously under the same lighting conditions during the secondary calibration process.
The present invention also provides apparatus for calibrating an imaging system, the apparatus including means for performing the operations or steps of the method of the invention.
It will be convenient to describe the steps of the method and apparatus in conjunction with each other since separating the descriptions of the method and apparatus will be artificial and may make understanding more difficult. The invention however provides both a method and apparatus.
The primary calibration of present invention enables compensation for characteristics of the camera or cameras used in different environments to ensure a system used in one factory, abattoir, etc will not provide different results to a different camera because of different cam era response characteristics. Preferably, the primary calibration is carried out at the comm
Barrett-Lennard David
Garvey Emlyn
Malmstrom Kurt
Partington Lachlan
Desir Jean W.
Drummond & Duckworth
Meat & Livestock Australia Limited
LandOfFree
Calibration of imaging systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Calibration of imaging systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Calibration of imaging systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3154305