Multispectral CameraSimon Belkin, Audrey Finken,
Grant George, Matthew WalczakFaculty Advisor: Prof. Mario Parente
Department of Electrical and Computer Engineering
ECE 415/ECE 416 – SENIOR DESIGN PROJECT 2013
College of Engineering - University of Massachusetts AmherstSDP13
Abstract
Block Diagram
System Overview
Results
Specifications
Acknowledgements
Team Logo
Multi-spectral cameras capture images through special optical filters, essentially band-pass filters, which allow only a certain range of wavelengths to pass through while blocking out the rest.
The multi-spectral camera system will be operated by a Raspberry-Pi brand microcontroller system. This system will position an individual filter in front of a monochromatic camera and an image will be captured, processed, and displayed to the operator
We would like to thank:• Seahorse Bioscience for donating the filter
wheel used in this project.• Akshaya Shanmugam for her help in
Spectrometer Testing
System electronically commands and controls the filter wheel assembly and monochrome camera
Differentiate between various rocks based on the spectrum provided
Pixels of images taken at various filters alignments Project budget is set at $500 Filter wheel thickness required optical engineering
to determine required focal lengths
The primary goal of the SDP13 Multi-Spectral Camera System is to develop an affordable multi-spectral imagery system that will have the ability to be installed onto a Mars Rover type of vehicle and perform imagery analysis at close to medium distances, (.3-10m). A secondary goal is to create a very useful, and relatively affordable multi-spectral imagery system that enables amateur scientists to view and learn about their surroundings in an affordable and non-complicated way
No Filter 425nm
436nm 450nm
860nm
510nm 750nm
990nm
Cost Accounting
Raspberry Pi
Getting Images via Image Registration
Development (R&D Costs)
Production Cost
Item Unit Cost Item Unit Cost
Filter Wheel $0.00 Filter Wheel $350.00
Raspberry Pi $0.00 Raspberry $35.00
Mightex USB Camera $219.00 Mightex USB Camera $127.02
Pentax Lens $99.95 Pentax Lens $57.97
Bi-Convex Lens $4.00 Bi-Convex Lens $2.32Adaptor - C-Mount to SM1 $19.75
Adaptor - C-Mount to SM1 $11.46
Lens Tube 0.5" $12.59 Lens Tube 0.5" $7.30
Lens Tube 3.0" $25.75 Lens Tube 3.0" $14.94
Stepper Motor Driver $14.95 Stepper Motor Driver $8.67Adaptor - C-Mount M/M $25.00
Adaptor - C-Mount M/M $14.50
Filter 425nm Filter $0.00 Filter 425nm Filter $57.42
Filter 436nm Filter $37.00 Filter 436nm Filter $20.00
Filter 670nm Filter $35.00 Filter 670nm Filter $20.30
Filter 750nm Filter $99.00 Filter 750nm Filter $57.42
Filter 860nm Filter $99.00 Filter 860nm Filter $57.42
Filter 990nm Filter $99.00 Filter 990nm Filter $57.42
Total Part Cost $785.99 Total Part Cost $899.16
Optics and Ray Tracing
Filter Selection
Above are the important wavelengths that differentiate the rocks based on those values
Pixel error = [1.654, 0.9432]Focal Length = (2034.19, 2087.65)Principal Point = (405.59, 402.928)Skew = 0Radial coefficients = (2.092, -15.64, 0)Tangential coefficients = (0.1224, -0.002745)
+/- [1811, 1886]+/- [74.77, 235.8]
+/- 0+/- [3.722, 54.88, 0]
+/- [0.2361, 0.03374]
0 100 200 300 400 500 600 700
0
50
100
150
200
250
300
350
400
450
5
55 5
5
5
10
10
10
1015
1520
25
Complete Distortion Model
X
YO
Image 2 - Image points (+) and reprojected grid points (o)
100 200 300 400 500 600 700
50
100
150
200
250
300
350
400
450
• Extension tubes added to facilitate a greater focal length.
• Thin lens equation, 1/di = 1/f – 1/do, determines the distance to the object, do, and to the image, di.
Image Registration is the process of estimating an optimal transformation between two images. We transformed a picture taken with filters based on a reference image taken without filters. The Software was done in python to work on the pi.
Pixel error = [1.654, 0.9432]Focal Length = (2034.19, 2087.65)Principal Point = (405.59, 402.928)Skew = 0Radial coefficients = (2.092, -15.64, 0)Tangential coefficients = (0.1224, -0.002745)
+/- [1811, 1886]+/- [74.77, 235.8]
+/- 0+/- [3.722, 54.88, 0]
+/- [0.2361, 0.03374]
0 100 200 300 400 500 600 700
0
50
100
150
200
250
300
350
400
450 5
55
10
10
10
10
15
15
15
15
20
20
20
20
25
25
25
30
3035
Radial Component of the Distortion Model
Pixel error = [1.654, 0.9432]Focal Length = (2034.19, 2087.65)Principal Point = (405.59, 402.928)Skew = 0Radial coefficients = (2.092, -15.64, 0)Tangential coefficients = (0.1224, -0.002745)
+/- [1811, 1886]+/- [74.77, 235.8]
+/- 0+/- [3.722, 54.88, 0]
+/- [0.2361, 0.03374]
0 100 200 300 400 500 600 700
0
50
100
150
200
250
300
350
400
450
5
55
5
10
10
10
10
15
15
15
20
20
20
25
25
2530
30353540
Tangential Component of the Distortion Model
Geometric calibration was done in Matlab to remove geometric distortions caused by various filters. The images above were taken from Matlab while performing camera calibration. It shows the distortion model of images from filters and the settings of the camera at various filters