Rethinking Image Histogram Matching for Image Classification

2025-06-02

AI Summary

Adverse weather conditions degrade image classification performance due to low-contrast inputs. To address this, we propose a differentiable, end-to-end parameterized histogram matching preprocessing method. Unlike conventional approaches that adopt fixed target distributions (e.g., uniform), our method models the target pixel distribution as a learnable variable—jointly optimizing its shape and the downstream CNN classifier’s loss—thereby tailoring the distribution specifically for classification rather than generic enhancement. The method leverages gradient-driven differentiable quantile mapping, enabling end-to-end training using only normal-weather images; no additional inference overhead is incurred. Evaluated across diverse adverse weather scenarios, our approach achieves an average 3.2% improvement in classification accuracy over baselines such as histogram equalization, significantly enhancing model robustness to weather-induced degradation.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Search and Optimization: Adversarial Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

This paper rethinks image histogram matching (HM) and proposes a differentiable and parametric HM preprocessing for a downstream classifier. Convolutional neural networks have demonstrated remarkable achievements in classification tasks. However, they often exhibit degraded performance on low-contrast images captured under adverse weather conditions. To maintain classifier performance under low-contrast images, histogram equalization (HE) is commonly used. HE is a special case of HM using a uniform distribution as a target pixel value distribution. In this paper, we focus on the shape of the target pixel value distribution. Compared to a uniform distribution, a single, well-designed distribution could have potential to improve the performance of the downstream classifier across various adverse weather conditions. Based on this hypothesis, we propose a differentiable and parametric HM that optimizes the target distribution using the loss function of the downstream classifier. This method addresses pixel value imbalances by transforming input images with arbitrary distributions into a target distribution optimized for the classifier. Our HM is trained on only normal weather images using the classifier. Experimental results show that a classifier trained with our proposed HM outperforms conventional preprocessing methods under adverse weather conditions.

Problem

• Improving classifier performance on low-contrast images
• Optimizing target pixel distribution for adverse weather conditions
• Differentiable parametric histogram matching for image preprocessing

Innovation

• Differentiable parametric histogram matching for classification
• Optimized target distribution via classifier loss function
• Training with normal weather images enhances adverse performance