Visit http://jethomson.wordpress.com/spectrometer-articles/ for a series of articles about this code. Visit http://jethomson.wordpress.com/spectrometer-articles/code to download sample spectrographs to process with this code. Before executing any of the scripts and functions you'll want to run setup_path. The octave packages octave-miscellaneous and octave-signal are required to run this code. This set of functions and scripts allows you to transform a spectrograph, a photograph of a light source's spectrum, into a spectrogram, a plot of the light's spectrum with a wavelength scale. Not all the steps listed here are necessary to produce a spectrogram. It is possible to produce a spectrogram with a wavelength scale quite simply from one spectrograph of a CFL, one spectrograph of the light source of interest, and a bit of processing with the functions wavelength_calibrate() and image2spectrum(). The spectrographs were taken by placing a diffraction grating over the camera's lens and photographing light after it passes through a narrow slit made from razor blades. The photos were shot with a custom white balance to prevent the white balance from changing as various spectrums are photographed. The spectrographs of the light spectrum of interest and the reference spectrographs must be taken in the same session to prevent changes in the environment from affecting the data. The scripts directory contains primary scripts and example scripts. The primary scripts either batch process spectrographs or setup the spectrometer code's workspace at the beginning of a new session. The example script example_02_cyan_LED_spectrogram_preprocessed.m shows the easy way to produce a spectrogram; however it hides away much of the inner workings so example01_cyan_LED_spectrogram_not_preprocessed.m shows more complicated examples of how the functions can be called and in what order. Before using the spectrometer functions and scripts, cd to the spectroradiometer directory and run setup_path.m ===Methods=== This section details the steps taken to produce a spectrograph and convert it to a spectrogram. Setup and Calibration Frames ---------------------------- 01) Let the camera warm up. So that the camera will be thermally stable when taking dark frames. 02) Set a custom, uniform white balance. (note 1) 03) Put on diffraction grating. Turn on CFL. 04) Using LCD view finder: position camera, rotate grating, and orient box to get a bright, horizontal spectrum. 05) Determine the best zoom and focus. 06) Use the lowest f-number (while zoomed) and lowest ISO. 07) Determine a shutter speed that will give a proper exposure for the CFL, incandescent, and light source of interest.* Use saturation_test.m to help. 08) Determine the best color of light to use for taking flats. Use find_camera_gray_test.m to help. 09) Determine the shutter speed that results in the flats' RGB histograms between 1/3 and 2/3 of max. 10) Take light darks using the shutter speed determined in step 7. 11) Take flat darks using the shutter speed determined in step 9. 12) Take flats. 13) Take bias frames. ** 14) Use batch_raw_decode.m to decode calibration frames. 15) Run make_master_dark_flat_bias.m *A light source can be moved closer to or further away from the slit to prevent under/overexposure. It is not necessary for all the light sources to be the same distance from the slit. However the height of every light source should be adjusted so that the center of it's spectrum occurs at half the camera's image height. **Bias frames are only necessary if you want to do dark frame matching. General ------- 01) Let the camera warm up. 02) Set a custom white balance. (note 1) 03) Put on diffraction grating. 04) Using LCD view finder: position camera, rotate grating, and orient box to get a bright, horizontal spectrum. 05) Set the camera to the settings determined in "Setup and Calibration Frames". 06) Take a CFL spectrograph. 07) Take an incandescent bulb spectrograph and call get_fO_data to take an irradiance reading. (note 2) 08) Take a spectrograph of the light source of interest and take an irradiance reading. 09) Follow blackbody steps to get Ti, Ri, and Rf. (note 2) 10) Copy the spectrographs and metadata into the proper directory structure. 11) Run batch_raw_decode.m 12) Run batch_preprocess.m (note 3) 13) Run spectrometer_setup.m 14) Edit the parameters used to determine the temperature of your incandescent calibration lamp. (note 2) 15) Run system_function_calibrated.m (note 2) 16) Use image2spectrum to produce a spectrogram from your spectrograph of interest. 17) Use Hr from system_function_calibrated to correct spectrogram of interest. (note 2) 18) Use an irradiance measurement and TSL230_fO_to_irradiance() to calibrate the spectrum of interest. (note 2) [1] A uniform white balance, where all the multipliers are equal (i.e. mr = mg = mb), can be obtained by setting a custom white balance on a target that produces an overexposed image. This won't work for all cameras. [2] This step is only required if you are interested in the spectrometer's system function and the effect it has on your spectrograph of interest. [3] If you are using an irradiance sensor affixed to one side of the slit, you need to determine what row number the center of the sensor corresponds to and use that row number to determine the region of interest you want batch_preprocess to crop the spectrographs to (i.e. y0 and h defined in batch_preprocess.m). An easy way to do this is to aim a laser pointer through the slit such that it is level with the center of the sensor, take a spectrograph, then examine it to find the row number. Blackbody --------- _No Power Applied_ 01) Make sure it's unplugged! 02) Measure temperature of room - Ti, temperature in Celsius. 03) Screw in short-circuit bulb base. 04) Measure Rx, closed loop resistance. 05) Screw in light bulb. 06) Measure R1, cold circuit's total resistance. 07) Measure In, meter AC noise current. 08) Measure Vn, meter AC noise voltage. 09) Make sure shunt is inactivated. _Power Applied_ 10) Measure V2, hot circuit's total AC voltage. 11) Attach current meter. 12) Shunt current through current meter.++ 13) Measure I2, hot cirucit's total AC current. _Calculations_ 14) Ri = R1 - Rx 15) Rf = ((V2-Vn)/(I2-In)) - Rx 16) Use calculate_filament_temp() to calculated the temperature of the hot filament. Tf = calculate_filament_temp(Ti, Ri, Rf) +Taking these measurements required a modified electrical outlet box that allowed an AC circuit to be interrupted, probed, and measured. If you are using a 100W bulb on a 120 Vac supply not much error will result if you assume Tf is 2800K. generate_Wbb_spectrum() will accept Tf instead of Ri and Rf. ++The resistance of a cold filament allows an inrush of current that may exceed the limits of the meter that is why a shunt is used to switch current through the meter after the filament has warmed up and it's resistance has increased. ===End Methods=== The following directory tree shows the basic directory structure the spectrometer functions and scripts expect. Not all files and folders are shown. Spectrographs are always saved in a directory named for the type of files it holds. JPG holds the JPEGs output by the camera and PGM holds the PGM files that result from developing the CHDK raw files stored in DNG. Only JPEG and PGM files are processed by this code. The directories eN (where N is a number) each store an ensemble of spectrographs, where an ensemble is a collection of multiple pictures of the same subject. For example, if one wanted to test two blue LEDs, e0 would hold an ensemble of spectrographs of the 1st LED and e1 would hold an ensemble of spectrographs of the 2nd LED. Reference spectrographs will never need more than one ensemble by virtue of them being references. The Hg folder holds a spectrograph of compact fluorescent lamp that is used to calibrate a wavelength scale. The W folder holds a spectrograph of an incadescent bulb that is used to determine the spectral response of the spectrometer. The dark, flat, and bias frames folders hold the images used to perform dark and flat frame corrections. data └── frames ├── bias_frames │ ├── DNG │ ├── JPG │ └── PGM ├── dark_frames │ ├── flat_darks │ │ ├── DNG │ │ ├── JPG │ │ └── PGM │ └── light_darks │ └── 2s │ ├── DNG │ ├── JPG │ └── PGM ├── flat_frames │ ├── DNG │ ├── JPG │ └── PGM └── light_frames └── 2s └── 2012_08_30 ├── reference │ ├── radiometric │ │ ├── DNG │ │ ├── JPG │ │ └── PGM │ └── spectral │ ├── DNG │ ├── JPG │ └── PGM └── spectrographs ├── 60W_GE_8 │ └── e0 │ ├── DNG │ ├── JPG │ └── PGM ├── 60W_Sylvania_8 │ └── e0 │ ├── DNG │ ├── JPG │ └── PGM └── Cyan_LED_3 └── e0 ├── DNG ├── JPG └── PGM ===How to Use the Code=== Most of the functions are setup in such a way that they will process individual images that have not been preprocessed. For example if you have an image of a CFL's spectrum named HG_REF.JPG and a spectrograph of an LED name CYAN01.JPG, then to plot the spectrum of the LED you can use the following code: lambda = wavelength_calibrate('HG_REF.JPG'); Z = image2spectrum('CYAN01.JPG'); plot(lambda, Z) The next step up in complexity is to take several spectrographs of the same light source to form an ensemble and to place the images in a special directory structure. batch_preprocess averages, crops, dark subtracts, and flat corrects all of the ensembles and saves them as mat files. Once this processing is done it doesn't need to be performed again. Dark subtraction, flat correction, and tone linearization are optional and can be turned off individually by uncommenting a couple of lines in batch_preprocess. Now that you have preprocessed the ensembles into mat files you can use the functions to process them much like you did with individual images. load('path/to/reference/spectral/PGM_spectral_ref.mat') lambda = wavelength_calibrate(spectral_ref); load('path/to/spectrographs/Cyan_LED_3/PGM_Cyan_LED_3_e0.mat'); Z = image2spectrum(spctgrph); plot(lambda, Z) To make things easier and keep the workspace tidy the scripts initialize_spectrometer_workspace and spectrometer_setup create a spectrometer object. The spectrometer object stores the wavelength scale and several other variables that make writing scripts to process spectrograms simpler. The script system_function_calibrated demonstrates how to use a incandescent bulb to calibrate the y-axis of a spectrogram in terms of spectral irradiance. It creates the so.Hr which when multiplied with a spectrogram converts counts to W/m^2/nm. Don't expect accurate radiometric calibration when using JPGs. Proper radiometric calibration requires that the ensembles be dark subtracted and flat corrected during the preprocessing step. Metadata -------- Metadata about a spectrograph should be saved in a text file that contains information such as distance from luminating surface to slit, irradiance measurements from the TSL230, the irradiance sensors sensitivity, units used, etc. batch_preprocess saves this metadata into the mat file it creates from each ensemble of spectrographs. Then when a spectrograph's mat file is loaded in a script its metadata can be used to radiometrically calibrate the resulting spectrogram.
jethomson/spectroradiometer
MATLAB/GNU Octave code for processing spectrographs produced with a point and shoot camera
MATLABNOASSERTION