IM Consultant Needed

[jwtrexler]

I am looking for an individual who can help writing command line script with the following project flow; for steps not best handled by IM, or handled in a better way, an output file needs to be formatted to manipulate the data in another program (e.g., Excel). At present, currently available IM scripts are available to preform the a number of the steps; steps I am needing assistance will be noted in capitalization

1) ACQUIRE DIGITAL IMAGE FROM CAMERA BY IM
There may be a script for this, as I know one exists for ImageJ, however I have not researched this yet with respect to IM.

2) Convert digital image from RGB to Cielab color space
At the present time, IM has a bug and cannot do the mathematical conversion to of an 8 bit, jpeg file. To accomplish step 2), I am converting the file from jpeg to tiff where IM is able to convert the file from 8 bit to 16 bit using the -depth script, then I use the -colorspace lab script to convert the file to L*a*b* space. One hurdle I'm running into is the Lab output is in percentages, so I write the tiff file to a txt file, import the txt file into Excel, parse the date removing excess characters for with respective pixels, and then have Excel calculate the "true/absolute" Lab value. Another limitation I am running up against is the fact that Excel can only handle slightly over one million lines of text and the images I am attempting to process have significantly greater than one million pixels. It may be possible to do some averaging in IM to overcome the limitation of Excel (or a functionally similar program), but this is another script that I have not identified if available.

3) OUTPUT FILE FORMAT IN A VISUAL FORMAT SUCH AS A HISTORGRAM
Finally with the image in a file comprising pixel versus Lab values, I would like to have the data displayed in a statistically useful format to determine if all images exhibit the same Lab colors.

What I am trying to develop is a program that will allow digital pictures to be taken of the top surface of 3 dimensional objects having a textured surface and being produced at a rate of part per 10 seconds. This subject IM program will be used to follow the production of the parts, allow an assessment of color for every part, and can generate an visual output (e.e., histogram) that can be utilized to insure all parts meet a color spec "live" as the part is produced. Further work may involve modifying the output from step 3) as more meaning visual formats are identified to determine whether part quality is meeting spec.

If you feel you have the qualifications to assist with this project and desire to have a more in depth discussion to determine if the project goal matches with your capabilities, interest, time, and compensation, please contact me at the number listed below.

Jack Trexler
(423) 292-4035

[fmw42]

1) As far as I know, IM has no direct way to access data from a camera. If the data is in a raw format, IM can read the data and convert to other formats if need. But it uses a delegate to read the raw data. I think that is DCRAW. But I am not an expert on this.

3) I have a unix bash shell script that uses GNUPLOT to convert an image into a graphical 256 bin histogram or graph a 1D (row or column) of an image into a profile. See my script, plot, at the link below.

Note in IM, LAB A and B channels are biased by 50%, since IM cannot deal with negative pixel values unless you compile IM in HDRI mode. However, txt: format is still reporting them zero-biased as percent. But that may change. See viewtopic.php?f=3&t=25830

I also have a histogram compare script, histcompare, that may be of interest.

I have used AWK to process the data from IM txt:. That seems to be comparatively quick compared to for loops and such, though not as good as having it coded in C or the like.

[Werty]

1) As far as I know, IM has no direct way to access data from a camera.

What about from an IP camera ?

I know IM can fetch images from the web by using a web address instead of a harddrive path´, so maybe an IP camera would too, dunno, never tried an IP camera, but maybe worth looking into.