i.his.rgb: reimplement conversion code

[jayneel-shah18]

This pull request fixes incorrect RGB outputs in i.his.rgb GRASS module for certain combinations of hue, intensity, and saturation values, particularly:

• When saturation = 0 (expected grayscale output)
• When intensity = 0 (expected black output)
• When intensity = 255 and saturation = 255 (expected pure hue)

The prior implementation failed to correctly compute RGB for these edge cases due to formula limitations, incorrect assumptions, and redundancy in the his2rgb.c logic.

This patch:

• Correctly computes pure grayscale and black cases
• Fixes hue interpolation logic with proper clamping and rounding
• Replaces the core conversion logic with a mathematically accurate implementation
• Preserves the rowbuffer overwrite style used in main.c

File Changed

• his2rgb.c — fixed all cases via modular hue2rgb() logic and improved edge-case handling.

Validation Table (Old vs. Reference RGB Values)

IDX	H	I	S	OLD_RGB	REF_RGB	Match	Justification
0	0	0	0	[0, 0, 0]	[0, 0, 0]	Yes	S=0: grayscale expected
1	0	0	128	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
2	0	0	255	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
3	0	128	0	[0, 0, 0]	[128,128,128]	No	S=0: grayscale expected
4	0	128	128	[192, 64, 64]	[192, 64, 64]	Yes	Match
5	0	128	255	[255, 1, 1]	[255, 1, 1]	Yes	Match
6	0	255	0	[0, 0, 0]	[255,255,255]	No	S=0: grayscale expected
7	0	255	128	[255,255,255]	[255,255,255]	Yes	Match
8	0	255	255	[255,255,255]	[255, 0, 0]	No	Expected pure hue
9	60	0	0	[0, 0, 0]	[0, 0, 0]	Yes	S=0: grayscale expected
10	60	0	128	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
11	60	0	255	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
12	60	128	0	[0, 0, 0]	[128,128,128]	No	S=0: grayscale expected
13	60	128	128	[139,192, 64]	[139,192, 64]	Yes	Match
14	60	128	255	[150,255, 1]	[150,255, 1]	Yes	Match
15	60	255	0	[0, 0, 0]	[255,255,255]	No	S=0: grayscale expected
16	60	255	128	[255,255,255]	[255,255,255]	Yes	Match
17	60	255	255	[255,255,255]	[0,255,0]	No	Expected pure hue
18	120	0	0	[0, 0, 0]	[0, 0, 0]	Yes	S=0: grayscale expected
19	120	0	128	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
20	120	0	255	[0, 0, 0]	[0, 0, 0]	Yes	I=0: black expected
21	120	128	0	[0, 0, 0]	[128,128,128]	No	S=0: grayscale expected
22	120	128	128	[64,192,169]	[64,192,169]	Yes	Match
23	120	128	255	[1,255,210]	[1,255,210]	Yes	Match
24	120	255	0	[0, 0, 0]	[255,255,255]	No	S=0: grayscale expected
25	120	255	128	[255,255,255]	[255,255,255]	Yes	Match
26	120	255	255	[255,255,255]	[0,255,0]	No	Expected pure hue
27	45	64	192	[107,112, 16]	[107,112, 16]	Yes	Match
28	90	128	64	[96,160,104]	[96,160,104]	Yes	Match
29	135	192	255	[129,233,255]	[129,233,255]	Yes	Match
30	180	128	128	[94,64,192]	[94,64,192]	Yes	Match
31	210	64	255	[120,0,128]	[120,0,128]	Yes	Match
32	240	192	64	[208,176,187]	[208,176,187]	Yes	Match
33	270	128	192	[224,100, 32]	[128,32,224]	No	Hue interpolation mismatch
34	300	255	128	[255,255,255]	[255,255,255]	Yes	Match
35	330	192	255	[159,255,129]	[255,129,192]	No	Hue interpolation mismatch
36	30	192	192	[239,212,145]	[239,212,145]	Yes	Match

Fixes https://github.com/OSGeo/grass/issues/5659

Please review and suggest further improvements if necessary.

[petrasovaa]

I realized that there are also tools r.his and d.his, could you check how different is that implementation?

[jayneel-shah18]

Thank you for the suggestion. I reviewed the implementations of r.his and d.his.

• r.his performs RGB to HIS conversion and shares the same theoretical model that I have used in this patch. Its hue interpolation logic is similar to the hue2rgb function in this patch for i.his.rgb.

• d.his is a display module designed for interactive screen rendering. Unlike i.his.rgb, it operates at the graphics level and does not use raster CELL buffers.

This patch complements r.his by implementing the inverse HIS to RGB transformation, using the same color space model and consistent logic.