Stackoverflow during complex math on android

Question

I am using a hough transform to find circles in an image on android. At once point I get a stack overflow:

12-03 16:18:20.999: I/dalvikvm-heap(21563): Grow heap (frag case) to 29.662MB for 2253716-byte allocation
12-03 16:18:21.014: I/dalvikvm(21563): threadid=11: stack overflow on call to Lorg/DTS/boltSizer/ImageProcessing/hystThresh;.hystConnect:VII
12-03 16:18:21.014: I/dalvikvm(21563):   method requires 44+20+12=76 bytes, fp is 0x5e949338 (56 left)
12-03 16:18:21.014: I/dalvikvm(21563):   expanding stack end (0x5e949300 to 0x5e949000)

Here is where the error is being thrown:

    private void hystConnect(int x, int y) {

        int value = 0;
        for (int x1=x-1;x1<=x+1;x1++) {
            for (int y1=y-1;y1<=y+1;y1++) {
                if ((x1 < width) & (y1 < height) & (x1 >= 0) & (y1 >= 0) & (x1 != x) & (y1 != y)) {
                    value = (input[y1*width+x1])  & 0xff;
                    if (value != 255) {
                        if (value >= lower) {
                            input[y1*width+x1] = 0xffffffff;
                            hystConnect(x1, y1);
                            ran++;

                        } 
                        else {
                            input[y1*width+x1] = 0xff000000;
                        }
                    }
                }
            }
        }

    }

I think I understand what it means, but how does one go avoiding this. If you need more of the code being ran just ask.

You couldn't find any better place to ask about your stackoverflow problems than this one :) — Marcin Orlowski, Dec 03 '12 at 22:30
for (int x1=x-1;x1<=x+1;x1++) this condition always satisfies isn't it? — kosa, Dec 03 '12 at 22:33
FOrgot to mention I have this working in an applet. Same exact code. Runs find on my PC but not on Android. Thats whats confusing me. — Aaron Decker, Dec 03 '12 at 22:34

score 0 · Answer 1 · edited May 23 '17 at 10:24

Start with this:

How to increase the Java stack size?

Android: Increase call stack size

Ultimately this just means your function is consuming a lot of memory and you will have to find a more efficient algorithm. This is similar to if you posted an O(e^n) function and asked how to make it go faster. If this is your situation then you will want to research a more space-efficient algorithm to use.

Stackoverflow during complex math on android

1 Answers1