"In the optical microscope, there is a basic trade-off between field of view and resolution: the finer, the smaller the area of the microscope imaging. A method of overcoming this limitation is to translate samples and acquire an image on a larger view. Basic thinking is to put many high-resolution images together to form large fov. In these images, you can see the complete samples, as well as the fine details of any part of the sample. The result is a consisting of about billion pixels. Image, much more than the photos taken with DSLR or smartphone, and the latter usually has approximately 10 million to 50 million pixels. View these Gigabit pixels to display large amounts of information in these images.
In this teaching, I will discuss how to build a microscope that can imaging 90mm x 60mm field, the pixels of the sample correspond to 2μm (although I think the resolution may approach 15 μm). The system uses a camera lens, but the same concept can be applied using a microscope objective to obtain a more fine resolution.
I uploaded my Gigabit pixel image with Microscope on Easyzoom:
National Geographic Magazine Image 1970
My wife made crochet tablecloth
Miscellaneous electronics
Step 1: Supply list
Material:
1. Nikon DSLR (I am using my Nikon D5000)
2. 28mm focal length lens, 52mm thread
3. 80mm focal length lens, 58mm thread
4. 52mm to 58mm reverse coupler
5. Tripod
6. Seven 3 mm thick plywood
7. Arduino nano
8. Two h bridges L9110
9. Two infrared transmitters
10. Two infrared receivers
11. Press the button
12. Two 2.2kOHM resistance
13. Two 150 ohms resistors
14. A 1kohm resistor
15. Remote release of Nikon cameras
16. Black poster board
17. Hardware kit
18. Two stepper motors (I use NEMA 17 bipolar stepping motor 3.5V 1A)
19. Two 2mm guide screws
20. Four pillows
21. Two screw nuts
22. Two bearing slip bushing and 200mm straight axis
23. 5V Power
24. Winding
tool:
Laser cutting machine
2. 3D printer
3. Allen wrenches
4. Shear clamp
5. Winding tool
Step 2: System Overview
Be
In order to transgate the sample, two stepper motors aligned in the orthogonal direction move in the X and Y directions. The motor is controlled by two H bridges and an Arduinoo. The IR sensor at the bottom of the stepper motor is used to zero all levels, so they do not enter any of the ends of the block. Digital microscopes are located above the XY platform.
Locate the sample and place the stage, press the button to start collecting. The motor moves the stage to the lower left corner to trigger the camera. The motor is then translated in a small step because the camera takes photos in each position.
After shooting all images, the image is then spliced together to form a gigabit pixel image.
Step 3: Microscope assembly
Be
I made a low-rate microscope with DSLR (Nikon 5000), Nikon 28mm F / 2.8 lens and Nikon 28-80mm zoom lens. The focal length of the zoom lens is set to 80 mm. The combination of these two lenses is like a microtaral lens and objective lens. The total magnification rate is the ratio of the focal length, about 3 times. These lenses are actually designed for this configuration, so in order to spread the photographic microscope, you must place a aperture between the two lenses.
First, install the lens of the longer focus to the camera. A circle is cut from the black poster board, which is substantially the same as the size of the front surface of the lens. Then cut a small circle in the middle (I have a diameter of about 3 mm). The size of the circle will determine the amount of light into the system, also known as numerical aperture (NA). For carefully designed microscopes, NA determines the horizontal resolution of the system. So why not use high NA in this setting? Then there are two main reasons. First, as the Na increases, the optical image difference of the system becomes more prominent and will restrict the resolution of the system. In such a non-conventional setting, the situation is likely to, thus increasing NA will eventually no longer help increase resolution. Second, the depth of field depends on NA. The higher the NA, the depth of field is shallow. This makes it difficult to focus all the non-flat objects. If NA becomes too high, you will only be limited to imaging microscope slides, which have thin samples.
The positioning of the aperture stop between the two lenses makes the system are roughly distinct. This means that the amplification rate of the system is independent of the sputum. This is important to bring images. If the object has a different depth, the view from two different locations will have an offset perspective (such as human vision). It is not challenging from a telephony imaging system that is unpredictable, especially in this high magnification.
The 28 mm lens is connected to an 80 mm lens using a 58 mm to 52 mm lens, and the aperture is located in the middle.
Step 4: XY Stage Design
Be
Be
I used the Fusion 360 to design the stage. For each scanning direction, there are four parts that require 3D printing: aphocgeter installation, two sliding unit expanders, and guide screw installations. The base and platform of the XY platform are cut by 3 mm thick plywood. The base fixing the X direction motor and the slider, the X platform fixes the Y direction motor and the slider, the Y platform fixed sample. The base consists of 3 sheets, and the two platforms consist of 2 sheets. In this step, a file for laser cutting and 3D printing is provided. After cutting and printing these components, you can follow up.
Step 5: Motor installation components
Be
Be
Use a winding tool to wrap the wire around the wires of two IR transmitters and two IR receivers. Color the wires in order to understand which end is the end. The lead wire on the diode is then cut, which is only started from the winding. The wire slides the wire over the motor space and then pushes the diode into place. The wire is guided until they leave the back of the unit to see them. These wires can be connected to the motor line. Now use four M3 bolts to install stepping motors. Repeat this step for the second motor.
Step 6: Stage assembly
Be
The base 1 and the base 2 are bonded together, one of which is a hexagonal opening for the M3 nut. After the glue is dried, the M3 nut is hammered into place. When the press is pressed, the nut does not rotate, so you can screw into the bolt in later. Now there is now the third backsheet (base 3) to cover the nut.
It is now time to assemble the lead nut mount. Clear any additional filaments from the bracket and push four M3 nuts into place. They are closely fit, so make sure to remove bolts and nut spaces using small screwdrivers. When the nut is aligned, push the guide nut into the base and secure it with 4 M3 bolts.
Install the bearing housing of the X direction linear converter, the slider mount, and the motor housing to the base. Put the guide nut assembly on the lead screw and then slide the guide screw into the position. Use a coupler to connect the motor to the lead screw. Put the slider unit into the rod and push the rod into the slider mount. Finally, the slider is used to install the extender using the M3 bolt.
The X1 and X2 plywood are glued together in a similar manner. The same process is repeated in the Y direction linear converter and the sample table.
Step 7: Scanner Electronics
Each stepper motor has four cables connected to the H bridge module. According to the above figure, the four cables of the IR transmitter and the receiver are connected to the resistor. The output of the receiver is connected to the analog inputs A0 and A1. Two H-bridge modules are connected to the pins 4-11 on Arduino Nano. The button is connected to the pin 2 via a 1 kohm resistor, which is convenient for user input.
Finally, the DSLR's trigger button is connected to the remote shutter, just like I do for the CT scanner (see step 7). Cut the remote shutter line. The wire mark is as follows:
Yellow - focus
Red - shutter
White - ground
For focusing lenses, the yellow line must be grounded. To take a photo, the yellow and red line must be grounded. I connect the diode and red cable to the pin 12, and then connect another diode and yellow cable to pin 13. Set as described in DIY Hacks and How-Tos Instructure.
Step 8: Get a gigabit pixel image
The attachment is the code of the gigabit pixel microscope. I use the Stepper library to control motors with H bridges. At the beginning of the code, you must specify the image of the microscope and the number of images you want to get in each direction.
For example, the microscope I made has a vision of about 8.2 mm × 5.5 mm. Therefore, I indicate that the motor moves 8mm in the X direction, shifts 5mm in the Y direction. 11 images are acquired in each direction, with a total of 121 images for complete gigabit pixel images (more details about this in step 11). Then, the code calculates the number of steps that the motor needs to be made in order to convert the platform according to this quantity.
How do you know where they are relative to the motor? How to translate in any end without hitting? In setup code, I have written a function that can move the stage in each direction until it breaks the path between the IR transmitter and the IR receiver. When the signal on the IR receiver is below a certain threshold, the motor stops. The code tracks the stage relative to the original position. Write the code so that the motor does not conversion too far, which will enable the platform into the other end of the pilot screw.
Once the stage is calibrated in each direction, the stage will be transferred to the center. Use a tripod, I put my DSLR microscope on the stage. Align the camera area with the intersection on the sample table. Once the stage is aligned with the camera, I record the stage with some painters, and then put the sample on the stage. Adjust the focus in the tripod z direction. The user then presses the button to start collecting. The stage is converted to the lower left corner and triggers the camera. Then the stage grating scan sample, and the camera takes a photo at each location.
There are also some code for troubleshooting motors and infrared sensors.
Step 9: Stitching image
Be
Once you have obtained all images, you are now facing challenges that put them together. One way to handle image splicing is to manually align all images in the graphics program (I use Autodesk graphics). This will definitely work, but it may be a painful process, and the edges of the image are obvious in a gigabit pixel image.
Another option is to use image processing technology to automatically sew together. This idea is to find a similar feature in the overlapping portion of adjacent images, and then transform the image to make the images align each other. Finally, by multiplying the overlapping part by linear weight factor and adding them together, the edge can be mixed together. If you are not familiar with image processing, this may be a daunting algorithm. I worked on the question for a while, but I couldn't get a complete and reliable result. This algorithm is best at a sample with very similar features, such as points in the magazine image. Attachment is the code I have written in Matlab, but it needs some job.
The last option is to use a Gigabit pixel photographic splicing program. I don't have any suggestions, but I know they are there.
Step 10: Microscope performance
In case you missed, the results are as follows: magazine pictures, crochet tablecloths and miscellaneous electronics.
The system specifications are listed in the above table. I tried to imaging with a 28mm and 50mm focal length lens. I estimate the best resolution of the system based on the diffraction limit (about 6 μm). It is actually difficult to test experimental testing without high resolution goals. I tried to print the vector file listed in this large-scale photography forum, but I am limited by the printer resolution. The best result of my printout is the resolution of the system <40 μm. I also look for small and isolated characteristics on the sample. The smallest feature in magazine printing is a ink point. I estimate is also about 40 μm, so I can't use it to better estimate the resolution. There are some small turf, very isolated. Because I know the field of view, I can calculate the number of pixels that occupy a small pit to estimate the resolution, about 10-15 μm.
In general, I am satisfied with the performance of the system, but if you want to try this project, I will have some precautions.
Stability of the stage: First, obtain high quality linear stage components. The components I use are much more than I think. I only use a slider installer in the suite for each rod, so this may be why the stage feels less stable. The stage is good for me, but this will become a problem for the system of higher magnification.
Higher resolution optical: The same idea can be used for a higher magnification microscope. However, a small motor having a finer step is required. For example, using this DSLR to enlarge 20 times will result in 1 mm field of view (if the microscope can image the large system without gap). ElectronUpdate uses a stepper motor in a CD player for a higher magnification microscope. Another trade-off will be shallow depth, which means that the imaging is limited to thin samples, and you will need a finer translation mechanism in the z direction.
Stability of tripod: The system can more stably use a more stable camera bracket. The lens system is very heavy, and the tripod is tilted from its design position 90 degrees. I have to press the foot of the tripod to help stabilize. The shutter can also sway the camera is enough to blur image. "
Our other product:
