Image analysis of surface texture

The attached image shows different surface pattern types. I've labelled tiger stripes as "1", leopard spots as "3" and amorphous noise as 2. I want to find filters that will enable me to automatically plot a heat map of the type attached, which is drawn by me by hand.

I have experimented with imgaborfilt, imgaussfilt and fibermetric. These can provide very localised metrics of surface pattern. But I really want to get a more coarse mapping of the type hand drawn by me. I realise that the borders will not be sharp like I've drawn. Any thoughts on how to do this?

Here's a code snippet that I've experimented with. It helps to identify patterns but I haven't been able to produce a map of texture types.

Ridge Detection & Metrics
%% =======================
bestThickness = 7;
bestSensitivity = 10;
ridgePercentile = 85;
densitySigma = 15;  % size of area-based smoothing (~30–50 px)
gaborWavelength = 8;  % matches center-to-center spacing
orientations = 0:10:170;

% 7a. Ridge (Hessian)
feat = fibermetric(img_adj, bestThickness, 'StructureSensitivity', bestSensitivity);
feat = mat2gray(feat);

% 7b. Ridge threshold
ridgeMask = feat > prctile(feat(binaryMask & specimenMask), ridgePercentile);

% 7c. Spatial density map
area_density = imgaussfilt(double(ridgeMask), densitySigma);
area_density = mat2gray(area_density);

% 7d. Gabor anisotropy
g = gabor(gaborWavelength, orientations);
gMag = imgaborfilt(img_adj, g);
aniso = std(gMag,0,3) ./ (mean(gMag,3)+eps);
aniso = mat2gray(aniso);

% 7e. Combined metric
combined_metric = area_density .* aniso;
combined_metric = mat2gray(combined_metric);