A fast infrared image enhancement algorithm based on contrast optimization model

基于对比度优化模型的红外图像快速增强算法

¹ College of Information Science and Engineering, Huaqiao University, Xiamen 361021, China
² Xiamen Multimedia Mobile Communication Laboratory, Xiamen 361021, China

Received: 23 November 2023

Abstract

In infrared image contrast enhancement, plateau histogram equalization is a fast algorithm with good performance, and the key lies in how to choose the appropriate plateau value. In order to address the shortcomings of existing plateau histogram algorithms and meet the needs of both high-performance and real-time processing, an optimization model for the triple plateau histogram algorithms based on contrast as an evaluation parameter is proposed. Firstly, in order to prevent excessive enhancement, the adaptive preprocessing is performed on the first plateau based on the image characteristics. Secondly, in order to achieve a balance between image enhancement and detail protection, a constraint criterion between the first and second plateaus was proposed. Then, in order to control the dynamic range of the resulting image, a third plateau value is set for the grayscale level that is in the bright area and has a probability density of zero. Finally, the present optimization model is applied to globally constrain the three plateau values, and the optimal plateau values are obtained through traversal optimization. Qualitative and quantitative experiments are conducted on several public databases, and the results show that the algorithm proposed in this paper has relatively better subjective results and objective metrics comparing with six existing plateau histogram algorithms. For 8 bit images, the processing time of the present algorithm is about 0.02 seconds, which has high real-time performance.

摘要

在红外图像对比度增强中, 平台直方图均衡是一种效果较好的快速算法, 关键在于如何选择合适的平台值。为了解决现有平台直方图算法的缺陷, 也为了能同时满足高性能与实时处理的需求, 针对三平台直方图算法, 提出了一种以对比度为评估参数的数学优化模型。根据图像特性对第1平台进行自适应预处理, 防止过度增强; 提出了第1平台与第2平台之间的约束准则, 达到图像增强与细节保护之间的平衡; 为处在亮区且概率密度为零的灰度级设置第3平台值, 以便控制结果图像的动态范围; 应用提出的优化模型对3个平台值进行整体约束, 通过遍历寻优得到最佳平台值。在多个公开数据库上进行了主观与客观实验, 结果表明, 与6种现有的平台直方图算法相比, 文中所提算法具有相对更好的主观效果与客观指标。对于8 bit图像, 文中算法的处理时间约为0.02 s, 具有较高的实时性。