Multispectral imaging measures color

Lighting, sensors, and image processors can affect any attempts to make absolute color measurements.

Lighting, sensors, and image processors can affect any attempts to make absolute color measurements. One of the purest forms of color measurement is computed by breaking down the visible spectrum into wavebands and assessing light intensities within each portion of the spectrum. Spectrometers perform this function using photodetectors. Multispectral imaging systems take this concept a step further, using CCD arrays to obtain waveband intensity measurements on a per-pixel basis across an entire field of view.

Al Piterman of Pixel Physics (Rochester, NY, USA; has designed an Imaging Colorimeter, an off-line inspection system, for a manufacturer of colored contact lenses. The system comprises a standard zoom microscope attached to a tunable VariSpec liquid-crystal filter (LCTF) from CRI (Woburn, MA, USA; The LCTF can be tuned in 0.1-nm increments across the visible spectrum (400 to 750 nm) through a serial cable connected to a PC host. A monochrome QImaging (Burnaby, BC, Canada; QI Cam with 1360 ¥ 1036 pixels and a macro video zoom lens from Thales-Optem (Fairport, NY, USA; with a close-up attachment acquires the light after it is filtered by the LCTF.

A 5400K light from Illumination Technology (East Syracuse, NY, USA; with an enhanced blue output helps to compensate for the LCTF's transmission in this area of the spectrum. FireWire (1394) output on the CCD camera alleviates the need for a frame grabber and allows the operator to import images directly into the proprietary software running on the host PC.

After the operator loads a contact-lens tray under the Imaging Colorimeter, the system begins acquiring images. Piterman used Image Pro Plus from Media Cybernetics (Silver Spring, MD, USA; to develop a program that takes 25 filtered images acquired by the QI Cam and combines their intensity and wavelength data into a single CIELab color space value for measurement against reference color data.

"Image Pro Plus was selected for our application because it provided a way to control the hardware, create a graphical user interface, and program image-processing routines," Piterman says.

"Visual Basic was used to create a graphical user interface for the spectral-acquisition software. The software-development cycle was shortened with the help of Auto Pro/IPBasic commands. It was easy to write a program that could control the hardware, perform automatic white-reference and spectral calibration, and calculate the resulting tristimulus values from the collected spectral image data," he adds.

More in Life Sciences