Flex-lbp improves texture recognition by learning customizable binary patterns
FLEX-LBP: Learning Local Binary Patterns with Exact Binary Inference for Texture Classification
Computer Vision and Pattern RecognitionMachine Learning
Summary
Texture recognition helps computers understand patterns in images, useful in fields like medical imaging and quality inspection. Traditional methods fix certain settings by hand, limiting adaptability. The authors created FLEX-LBP, a way to teach the computer these settings automatically while keeping the results easy to understand and fast to compute. They showed FLEX-LBP performs better than common texture analysis methods across various color texture datasets.
What this means in practice
- •For medical imaging technicians: Accurately classify tissue types in histology images using an improved texture descriptor that adapts its parameters to the dataset.
- •For industrial quality inspectors: Use learned binary texture patterns to better detect surface defects in manufacturing with fast and interpretable descriptors.
Authors
Mohamed Alimoussa, Amal El Kaid
Abstract
Local Binary Patterns (LBP) remain among the most widely used texture descriptors, from industrial inspection to medical imaging. Yet their radius, sampling angles, color transform and thresholds are fixed by hand for each dataset. We present FLEX-LBP, which learns these parameters by backpropagation while every forward pass computes exact binary codes and hard histogram counts. A straight-through estimator acts only in the backward pass, so the deployed descriptor is an interpretable hard-binary LBP with 90 learnable parameters and unchanged inference cost. In a controlled ablation on six color texture benchmarks, the component worth learning is dataset-dependent. One configuration, applied unchanged everywhere, learns the geometry and thresholds with the color transform frozen. It improves the nearest-neighbor accuracy of the identical fixed descriptor by 3.55 points on average over five seeds, significantly on five of six benchmarks and never significantly worse. With a linear discriminant readout, it gains 0.9 points; the fully learned descriptor gains 2.7 and has the highest mean. FLEX-LBP is more accurate than five classical LBP-family descriptors, PCANet and a neural LBP on five of six benchmarks. Its mean nearest-neighbor accuracy exceeds that of frozen ImageNet ResNet-50 features (23.5 million parameters). On the Kather colorectal cancer histology benchmark, the fully learned descriptor reaches 93.8%, significantly above the best classical descriptor (Completed LBP, 92.8%). The learned radii adapt in directions that recur across seeds. Noise drowns single-pixel comparisons and illuminant change flips inter-channel ones; median sampling or gray-world normalization restores 96-97%. The code is available as a drop-in PyTorch layer at https://github.com/medalimoussa/FLEX-LBP.