水下视觉SLAM图像增强研究

Underwater Visual SLAM Image Enhancement Research

针对水下图像质量降低导致视觉SLAM中图像特征点提取与匹配数量减少的情况,提出一种采用图像增强算法改善对比度提高图像特征匹配数量的方法。本文分析了水下图像在水体吸收和散射作用下导致图像质量下降,以及由此造成的视觉SLAM定位精度降低的原因。对比了空域和频域图像增强算法,选择直方图均衡化作为图像预处理方法,研究了直方图均衡化过程。在实际海域分别采集了对比度不同的两组图像,利用ORB特征法进行特征提取与匹配。实验表明,图像增强对水下图像特征点提取、匹配数量以及正确匹配数量均具有较大改善,特别是对比度不高的图像,效果更为明显。

Against the reduced number of feature points extracted and matched in visual SLAM due to the degradation of underwater image quality,a method using image enhancement algorithms to improve the contrast and increase the number of feature matches in the image was proposed.This paper analyzes the reasons for the degradation of the image quality caused by the absorption and scattering of water in the underwater image,and the reasons for the decline in the accuracy of the visual SLAM positioning.The spatial and frequency domain image enhancement algorithms are compared.Histogram equalization is selected as the image preprocessing method,and the histogram equalization process is studied.Two sets of images with different contrast were collected in the actual sea area,and the features were extracted and matched using the ORB feature method.Experiments show that image enhancement has greatly improved the feature point extraction,the number of matches,and the number of correct matches in underwater images,especially for images with low contrast.