How to caculate ORB match rate of two image

OpenMV Products
1

Hi,
Currently I need to get the match rate of two image, and try to get ORB descriptor’s key point size.It seems no method to get the size.Any ideas?
Thanks in advance.

2

Looks like the object type wasn’t built out. Um, if you save the descriptor to disk then you can read it back in as a binary file in the mean time to get this info

Here’s some code showing the format:

3

First 4 bytes are the type. Second 4 bytes are the number of keypoints (size).

4

Thanks, I have change the py_image.c and it is work.
The code is below:

mp_obj_t py_kp_size(mp_obj_t self_in)
{
	return mp_obj_new_int(array_length(((py_kp_obj_t *)self_in)->kpts));
}

STATIC MP_DEFINE_CONST_FUN_OBJ_1(py_kp_size_obj, py_kp_size);
STATIC const mp_rom_map_elem_t py_kp_locals_dict_table[] = {
    { MP_ROM_QSTR(MP_QSTR_size), MP_ROM_PTR(&py_kp_size_obj) },
};

STATIC MP_DEFINE_CONST_DICT(py_kp_locals_dict, py_kp_locals_dict_table);
static void py_kp_print(const mp_print_t *print, mp_obj_t self_in, mp_print_kind_t kind)
{
    py_kp_obj_t *self = self_in;
    mp_printf(print, "size:%d threshold:%d normalized:%d", array_length(self->kpts), self->threshold, self->normalized);
}

static const mp_obj_type_t py_kp_type = {
    { &mp_type_type },
    .name  = MP_QSTR_kp_desc,
    .print = py_kp_print,
    .locals_dict = (mp_obj_t) &py_kp_locals_dict,
};
5

Hi,

I’ll make a class for the match object and a class for the keypoints descriptor in the next release, thanks for bringing that up, also if you have any feedback/question on ORB/Keypoints please let me know. Thanks!