This project aims to count a number of (by default blue) dots on an image. It was originally built for counting amounts of cells on breeding grounds in order to simplify the tasks of e.g. chemists and biologists.
| Raw image | Selected points |
|---|---|
 |
 |
Dots outlined in green are recognized as one Dots. Yellow outlined dots were detected as a cluster of dots. The actual number of dots counted is determined by the area. If the area is twice as large as an average dot, two dots are counted for this.
To use this application a python of version 3.9.7 (https://www.python.org/downloads/) or higher is recommended. Additionally, the following packages are required:
- cv2
- tkinter
- tqdm
- numpy
- pillow
To install these packages simply copy the following command into your command line
pip install -r requirements.txt
or execute the script installRequirements.bat. This script automatically installs all required packages.
To pull the project using git type the following line:
git clone https://github.com/FabianSieper/countDots
To run the GUI version of the program, enter the following line into your command line:
python app.py
or double click the file app.py.
You can modify contrast and saturation by sliding the corresponding slider.
- Contrast
- Set the amount of contrast applied to the image
- Saturation
- Set the amount of saturation applied to the (contrast) image
- This setting is applied after the contrast of the image has been adjusted
- Brush size
- Set the size of the brush
- Smallest dots
- Set the minimal dot size
- Biggest dots
- Set the maximal dot size
- Max Dot collection-size
- Set the maximal size, a collection of dots can have
There are three different previews, which can accessed:
- Contrast boost
- Displays the original selected image, which has an increase in contrast. For this click on the button named
ShowContrImg.
- Displays the original selected image, which has an increase in contrast. For this click on the button named
- Saturation boost
- Displays the contrast-increased image in addition with an increase in saturation. For this click on the button named
ShowSatImg.
- Displays the contrast-increased image in addition with an increase in saturation. For this click on the button named
- Final image
- Displays the final image, containing the detected dots, the amount of detected dots and further possibilities. For further possibilities read the section Usage (Raw). For this click on the button named
Compute.
- Displays the final image, containing the detected dots, the amount of detected dots and further possibilities. For further possibilities read the section Usage (Raw). For this click on the button named
- Previous Image
- This buttons only appears, if you choose to process multiple files inside a folder. The button iterates backward through the images inside of the folder.
- Next Image
- This buttons only appears, if you choose to process multiple files inside a folder. The button iterates forward through the images inside of the folder.
- Save Final Image
- This button saves the final image, which is displayed by pressing the button Compute. The files are by default stored in the processedFiles folder inside the project directory.
- Reset
- When selecting an area of interest, mistakes can be made. Pressing the button Reset resets the selected area and reloads the original image.
If you are happy with your selected settings, you can save the final image by pressing the button SaveFinalImage.
Use the mouse to narrow down the area of interest by overdrawing areas. Points are not searched in overdrawn areas.
If a mistake was made while selecting an area, pressing the button Reset reloads the original image and removes selected areas.
Even more settings are available. For further information lookup descriptions in section Usage (Raw).
To run the raw scripts enter the following line into your command line:
python main.py
or double click the file main.py. This requires python to be selected as the default application for files with the ending .py.
You are asked to select whether you'd like to select a single file or multiple files inside of a folder (type "file" to select a single file). The selected file(s) are then processed, the amount of found dots displayed in the command line and an image (or multiple images) are saved under the folder processedFiles/. These images display which dots were found (green), which dots seem to be merged (yellow) and which areas aren't detected as dots (red).
For the best performance over multiple images, the same distance to the dots should be maintained. This is, because the area detected is depending on the resolution of the dots, which varies when the distance varies.
All settings can be found in the file code/settings.json.
As the size of dots (measured in area, amount of pixels) might differ between cameras, they can be changed. For this adjust the variables s1, s2, s3 to your liking.
s1: The minimal dot size. If a dot's area is smaller than s1, it won't be counted.s2: The maximal dot size. If a dot is between s1 and s2, the dots are detected. If they are bigger than s2, they aren't detected as single but as multiple dots.s3: The maximal area a collection of dots can have. If an area is bigger than s2, the area might represent multiple dots, laying near each other. If the area is bigger than s2 and also smaller than s3, the size of the area is divided by the average dot size (s1 + s2) / 2. By doing so, an approximate amount of dots in this area can be calculated.
- green contours: A single detected dot
- orange contours: Multiple detected dots
- red contours: No dots detected, as the area is too big to be a bundle of dots
If some of the dots are too close to each other, it can appear that they may be a single dot. For this, the area found is compared to the maximal set dot size. If the area is not too large, the area is divided by the maximal dot-area. By this, an approximate amount of dots is calculated. To show how many dots are approximated per found instance, set the parameter show underneath the parameter amountDotsCounted to true.
By default only dots with the color blue are minded. To change the color range, you can simply modify the values lowerColor and upperColor in the settings. The color values are set in the hsv-color space. Thus the values have to be:
Color = [Hue, Saturation, Value]
with the max values being
Color = [180, 255, 255]
If you transfer color codes from other applications or websites, keep in mind the scale them accordingly.
If the image is washed out, or the dots might not be as distinguishable from each other, you can increase the contrast of the image. Settings regarding the contrast can be found underneath the parameter contrast.
contrastIncrease: A number describing how much the contrast shall be increased.- <1 describes a reduction in contrast
- =1 has no effect on the contrast - it stays the same
- >1 increases the contrast
showImage: Describes, if the saturated image shall be shown. This option can be used in order to calibrate the program.
By default the saturation increase has the value 1 and no image regarding the contrast is shown.
As some dots might not be as colorful as others, the option for boosting saturation exists. Settings regarding the saturation can be found underneath the parameter saturation. This option builds up on the settings contrast. At first the contrast is applied.
saturationIncrease: How much shall the saturation be increased. The values 0 would describe no increase in saturation.showImage: Describes, if the saturated image shall be shown. This option can be used to calibrate the program.
By default, the saturation increase has the value 10 and no saturated increased image is shown.
If you'd like to compute multiple files at once, you can place all your files inside of one folder. All files inside of this folder will be processed. For this, change the parameter source to folder. By default, only one file can be selected. (Available parameter: file and folder).
To automatically view the final image, which contains marked dots, set the parameter showFinalImage to true. By default, the value false is set. However, it should be kept in mind, that showing the image at the end lowers the performance of the program.
In addition to Show final image it is also possible to overlay the original image with the image containing the marked dots. To enable this adjust the parameter overlayOriginalImage. The possible values range between 0.0 and 1.0, where 0.0 describes full transparency of the original image and 1.0 no transparency. A value of 0.8 is recommended if wanted. By default, the value is 0.0.
If you'd like to save the image, set the parameter saveImage to true. This also represents the default value. If no image shall be saved, set the value to false.
To change the folder where the final images will be saved, adjust the parameter saveDir. This parameter has to contain a path to a folder. By default, the images are saved inside the folder processedFiles/, which again lies inside the project folder.
If you'd like to display the amount of dots counted on an image, you can enter the section countedDotsToImage inside of the settings-file. These parameters all belong to the text for displaying the amount of dots counted on an image.
show: Shall the text be shown?lineWidth: How thick shall the text be?color: Which color shall the text be in?position: At which position shall the text be? [x, y]fontSize: How big shall the text be?
If you'd like to display the name of a file on the image itself, you can enter the section fileNameToImage inside of the settings-file. These parameters all belong to the text for displaying the name of a file.
show: Shall the text be shown?lineWidth: How thick shall the text be?color: Which color shall the text be in?position: At which position shall the text be? [x, y]fontSize: How big shall the text be?
If you are using the GUI-variant of the program the possibility of adjusting the windows size if given. For this adjust the parameter canvasSize to your liking. This parameter describes the maximal length and width of the canvas. The images are scaled accordingly.
If you are using the GUI-variant of the program you have to possibility to change the size of the brush. The brush is used to select parts of the image, which shall not be process while searching dots.