Help: Evaluate surrounding pixels in a radius

Programming Questions

crapitect..

Help: Evaluate surrounding pixels in a radius

in Programming Questions • 2 years ago

After a twenty year hiatus from programming, I'm jumping back into the mix and need a little help. I've done a lot of searching on these forums and the web trying to find something close that could give me a direction, but I'm stumped.

I'm trying to write a sketch that will search an image for green dots of a known radius and create a symbol in the center of those dots. Eventually it will loop to search for other colors and spawn other symbols.

My problem is figuring out how to evaluate the surrounding pixels based on a radius rather than a box matrix.

Thanks for the help.

Heres the code so far:

PImage img_existing;
void setup() {
size (200,200);
background (70);
img_existing = loadImage("find_test.png"); //200,220 image with dots
}
void draw() {
// define what color to search for
search_color = color(0,255,0)
//Call function to find the dots
Find_Existing(img_existing, 10, search_color);
}
// Function for finding the existing dots
void Find_Existing(PImage _img, int dot_radius, color(_search_color)) {
//Load the pixels of the screen
_img.loadPixels();
//loop through each pixel
for (int y = 0; i < height; y++) {
for (int x = 0; x < width; x++) {
//location in the pixel array is determined by
int loc = x + y*width;
// pull out the rgb values for the pixels
float r = red (_img.pixels [loc];
float g = green (_img.pixels [loc];
float b = blue (_img.pixels [loc];
/*
The radius check: if the surrounding pixels in the given radius around
the pixel being evaluated is equal to the specified color then spawn
an ellipse.
This is where I need help... not with the ellipse though ;)
*/
}
}
}

Replies(4)

PhiLho

Re: Help: Evaluate surrounding pixels in a radius

2 years ago

Well, a way to do this is to check the pixels in a... box around the given pixel. But if the dist() of a pixel of the box from its center is above the radius (the half size of the box), just don't check it.
Since dist() is costly, if you want speed, you might want to create an array (corresponding to the box) of booleans (too far = false, don't check, in range = true, check).

crapitect..

another approach to the underlying problem: parse svg file for elements

2 years ago

@phi.lho: thanks for the advice!

After struggling to find a good solution for all cases like overlapping dots. I turned to using an SVG as a source file rather than a raster image. Using geomerative tools it was much easier to find the center of the elements.

Below is my solution in case it will help others. Any tips on making it more elegant, I'd love to hear them.

RShape[] parseSVG(String ename) { // THE FUNCTION DEFINITION : RShape[] is the return type - an RShape array
child = new RShape [1]; // initializes an instance of the child array for use
/////// Start Parsing
int i=0;
// begin the loop to fill the array
for (i = 0; i < child_count; i++) {
String searchee = svg.children[i].name; // initialize a string to hold the name of child
int index = searchee.indexOf(ename); // initialize the the index of our search, if it is == 0 then we know we've found the child we are looking for
if (index != 0 && i < child_count) { // if index isn't 0 (false) these aren't the droids we're looking for, search again
}
else { // else means we found what we are looking for now do this
if (child[0] == null) { // check to see if our first slot in the array is empty
child[0] = svg.getChild(svg.children[i].name); // if it is, put a child here
}
else { // if not...
child = (RShape[]) append(child,svg.getChild(svg.children[i].name)); //add the child to the array
}
}
}
return child; //send back the filled array
}

archi_stu..

Re: Help: Evaluate surrounding pixels in a radius

2 years ago

have you gotten this to work? i dont understand how the second portion of code could be used to make it work. i am also needing help on tree recognition script

PhiLho

Re: Re: Help: Evaluate surrounding pixels in a radius

2 years ago

In the second part, the OP avoided the problem of image analysis instead of solving it...

Top Reply