nx_img_segment.nx source
↩ module page · 74 lines · 3861 B
1// nx_img_segment.nx -- R3 of the sovereign scanning stack: binary image -> connected components (glyph boxes).
2// 8-connected component labeling by iterative flood-fill (explicit stack, no recursion), accumulating each
3// component's bounding box as it grows. Ink = 0, paper = 255 (the R2 binarization convention). Each component is a
4// candidate glyph/mark that R4 (OCR classify) will recognize. Integer-exact, zero deps. license_tier: ORIGINAL
5import "nx_syscalls.nx"
6import "nx_img_threshold.nx" // -> nx_img_core (pixel buffer, decode, Otsu) for end-to-end gating
7
8const SEG_BOXF: i64 = 4 // per box: [minx, miny, maxx, maxy]
9
10// label 8-connected ink components; fill labels[] (0 = paper/background, >=1 = component id) and boxes[] (bbox per
11// component, index id-1). Returns the component count (capped at maxcomps). npix scratch stack allocated internally.
12func seg_components(pix: *u8, w: i64, h: i64, labels: *i64, boxes: *i64, maxcomps: i64) -> i64 {
13 let npix: i64 = w * h
14 var i: i64 = 0
15 while i < npix { labels[i] = 0; i = i + 1 }
16 let stack: *i64 = sys_mmap((npix + 1) * 8) as *i64
17 var ncomp: i64 = 0
18 var p: i64 = 0
19 while p < npix {
20 if pix[p] == (0 as u8) { if labels[p] == 0 {
21 ncomp = ncomp + 1
22 if ncomp > maxcomps { return ncomp }
23 let id: i64 = ncomp
24 let cx0: i64 = p % w; let cy0: i64 = p / w
25 boxes[(id-1)*SEG_BOXF+0] = cx0
26 boxes[(id-1)*SEG_BOXF+1] = cy0
27 boxes[(id-1)*SEG_BOXF+2] = cx0
28 boxes[(id-1)*SEG_BOXF+3] = cy0
29 var sp: i64 = 0
30 stack[sp] = p; sp = sp + 1
31 labels[p] = id
32 while sp > 0 {
33 sp = sp - 1
34 let cur: i64 = stack[sp]
35 let cx: i64 = cur % w; let cy: i64 = cur / w
36 var dy: i64 = 0 - 1
37 while dy <= 1 {
38 var dx: i64 = 0 - 1
39 while dx <= 1 {
40 var is_self: i64 = 0
41 if dx == 0 { if dy == 0 { is_self = 1 } }
42 if is_self == 0 {
43 let nxp: i64 = cx + dx
44 let nyp: i64 = cy + dy
45 if nxp >= 0 { if nxp < w { if nyp >= 0 { if nyp < h {
46 let np: i64 = nyp * w + nxp
47 if pix[np] == (0 as u8) { if labels[np] == 0 {
48 labels[np] = id
49 stack[sp] = np; sp = sp + 1
50 if nxp < boxes[(id-1)*SEG_BOXF+0] { boxes[(id-1)*SEG_BOXF+0] = nxp }
51 if nyp < boxes[(id-1)*SEG_BOXF+1] { boxes[(id-1)*SEG_BOXF+1] = nyp }
52 if nxp > boxes[(id-1)*SEG_BOXF+2] { boxes[(id-1)*SEG_BOXF+2] = nxp }
53 if nyp > boxes[(id-1)*SEG_BOXF+3] { boxes[(id-1)*SEG_BOXF+3] = nyp }
54 } }
55 } } } }
56 }
57 dx = dx + 1
58 }
59 dy = dy + 1
60 }
61 }
62 } }
63 p = p + 1
64 }
65 return ncomp
66}
67
68// bounding-box accessors (id is 1-indexed).
69func seg_box_minx(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+0] }
70func seg_box_miny(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+1] }
71func seg_box_maxx(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+2] }
72func seg_box_maxy(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+3] }
73func seg_box_width(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+2] - boxes[(id-1)*SEG_BOXF+0] + 1 }
74func seg_box_height(boxes: *i64, id: i64) -> i64 { return boxes[(id-1)*SEG_BOXF+3] - boxes[(id-1)*SEG_BOXF+1] + 1 }