Mark license plate area in images
$30-250 USD
Paid on delivery
The project is very simple, u have to extract license plate area of the car.
You don't have to do text recognition, just the extraction of the plate.
In Malaysia, license plates are black and this project is from Malaysia.
Since the license plates are black, the text inside the plate is a light color (mostly white).
Below is one approach that I recommend being a coder myself:
One method could be to extract all the pixels whose color is black or near black, and discard all other colors or pixels in the image. Whether the color of a pixel is black or near black can be recognized as below:
Say RBG of the color is (X, Y, Z) coordinates.
Black is (0,0,0). Distance of a color (X,Y,Z) from black (0,0,0) is sqrt((X-0)^2+(Y-0)^2+(Z-0)^2) or sqrt(X^2+Y^2+Z^2). Now if this distance is smaller than a threshold (t, say 50). Further, if this doesn't turn out to be accurate model, u can further incorporate another limit that Abs(X-Y), Abs(Y-Z), Abs(Z-X) should be less than some threshold, d, to give it a tendency of choosing either or near black or near grey.
Once all the dark pixels are extracted, make a vertical line at the center of the image. Loop through all the chosen dark pixels that fall on that line. Now some of these pixels are definitely going to be a part of the plate. Follow connected component labeling algo on each pixel to recursively mark all the pixels that are connected directly or indirectly to the current pixel in the loop. Once connected components have been performed, one of the components out of all will be the plate, which can be identified easily through some trivial logic (I leave up to you to figure out, but u can ask me if u can't figure out).
However, you are free to use your own methods, the above was just a suggestion of how I would do it.
Below is the description directly from the client:
Write a python script to process an image of cars and extract the license plate.
Text recognition is not needed, only extraction of the rectangular area of the license plate.
Following algorithms could be helpful to implement.
(Below techniques are only suggested by me, but you can use any method/algos or combination of algos. In the end, you will end up with an algorithm around histogram algo)
1-grey scale - [url removed, login to view]
2-binary - [url removed, login to view]
3-erosion - [url removed, login to view]
4-dialation - [url removed, login to view]
5-edge detection - [url removed, login to view]
6-histogram - [url removed, login to view]
And you may use PIL, numpy, matplot, opencv libraries.
Whoever bids $150 or less first will be awarded the project right away. Only if none of the bidders bid $150 or less, then I will increase the price.
Project ID: #6786168
About the project
11 freelancers are bidding on average $176 for this job
I am an expert in delivering custom scripts and tools willing to discuss further about the project needs.
Hello, Can you send me a sample image of the license. It is easy for me to do, But i want to make sure it is all you need :) Please send me the image that needs to extract license plate area of the car
Hi! I am professional C/C++/C#/Java programmer. I can do this proiject with highest quality! Best Regards, Szymszteinsl
Hi. I work in area of applied mathematics. My primary areas are programming languages, machine learning and digital signal processing. All research work I doing with MatLab. I write in C/C++, Python and Java most More
I did this project before Please send me the sample images to give you the plan for this project. I have been working in opencv for 8 years. Please check my profle for more information
*** Please check my portfolio *** Hi, Dear Sir. "High Quality + Powerful Service + Rapid Communication!" is my forever business policy! I just read your job posting with interest and carefully, and I’ll waiting inte More
I am a researcher in Image Processing. I have made a podium finish in an international competition twice. I have over three years of experience with opencv. I also have a research paper on image segmentation and ha More
