Hot Goal Detection with Red Light

[rebelalliancep]

I was wondering what ranges need to be used for the thresholdHSV command used in autonomous. We are currently using a red LED surrounding our camera. I noticed that there was a way to get the range using a the NI Vision Assistant, but our copy of it has already expired. Thanks in advance.

[mwtidd]

We used a blue light, and an RGB threshold. With the exposure cranked way down on our axis camera, we had no issues detecting hot at week zero.

HSV is great for complex vision processing, but is really unnecessary if you are using the reflective tap and an LED ring light. I would expect that your G and B values would be very low.

For the blue light we had the following rgb values:

int redLow = 0;
int redHigh = 10;
int greenLow = 0;
int greenHigh = 10;
int blueLow = 50;
int blueHigh = 110;

We didn't need to do any calibrations at week zero.

You should be able to load your image into any photo editing tool and measure the rgb values. Gimp provides both RGB and HSV. I just find RGB more intuitive.

[Greg McKaskle]

If the ranges of numbers are small, RGB is fine. HSV or HSL or HSI really come in handy when you are looking for a larger range. RGB can only describe rectangles. HS? can describe doughnuts, circles, triangles, and remain true to your intended color.

You should be able to install the newer Vision Assistant from the DVD.

Greg McKaskle

[otherguy]

One of the students on our team last year, or the year prior came up with a pretty handy solution.

Open the live feed for the camera in the web browser, aim it at a target have the ring light turned on.

Then use an application like Just Color Picker. No installer just an exe. It will run on top of the browser which has the video feed in it. Just place your mouse cursor over the live image and record the min/max values for the RGB values that you see in your image. You may need to widen up the ranges to support different lighting conditions.

Your camera should be calibrated prior to running through this sequence.

[Greg McKaskle]

The default dashboard logs one image every second to the logged in user's home directory/My Documents/LabVIEW Data. The images are named 00.jpg to 59.jpg.

These images can be loaded into any package you like, including Vision Assistant, and analyzed to review color, focus, positioning, lighting, shake, cables and other mechanisms obscuring the image, etc.

Greg McKaskle