ndrd / fast-edge Goto Github PK

line 295 of fast-edge.c

if dir[x + y] is 0 in the first line of the image, then y is zero,
so then g[x + y - w] is out of bounds --> possible segfault.
similarly for y = max_y - w, then g[x + y + w] is out of bounds.

it looks like 3 of those cases can go out of the array bounds either before or 
after, so they need some bounds-checking.

I think there's a small bug in the trace() function.

In fast-edge.c on line 422:
if (!(y == 0 && x_off == 0) && ...

should be:
if (!(y_off == 0 && x_off == 0) && ...

Cheers!

What steps will reproduce the problem?
1. non_max_suppression
2.
3.

What is the expected output? What do you see instead?
I see "ERROR - direction outside range 0 to 3"

What version of the product are you using? On what operating system?
eclipse + TDM-GCC-64 on Win 7

Please provide any additional information below.
In the non_max_suppression step, 
the switch (dir[x + y]) {
:
case 0:
case 1:
case 2:
case 3:
:
}

the variable, dir[x + y] = g[x + y] = g[x + y - w] = g[x + y] = g[x + y + w] = 
-1163005939, is that normal?
Thank you!

What steps will reproduce the problem?
1. perform test with hight image resolution ex. 3000x1024
2. work fine with low resolution 512x512... 
3. off

What is the expected output? What do you see instead?
segmentation fault

What version of the product are you using? On what operating system?
GCC +win 7

Please provide any additional information below.
==> crash at the 1, or 2 iteration of the first for loop:
for (y = 0; y < max_y; y += w) {//crash after one or two loop

-lm is required to avoid a linker error