With the development of computer vision, the semantic segmentation of remote sensing images, which has become an important topic, has been utilized in various applications for image content analysis and understanding, such as urban planning, natural disaster monitoring, and land resource management. Many approaches have been proposed to address these problems. However, due to obvious differences in resolution, spatial structure, and semantics between remote sensing images and ordinary images, the semantic segmentation of remote sensing images is still challenging. In this paper, we propose a novel multiscale image generation network (MIGN) that can efficiently generate high-resolution segmentation results by considering both details and boundary information. In particular, a multi-attention mechanism method for semantic segmentation of remote sensing images is designed. The attention weight is calculated by capturing the interaction of cross dimensions in a two-branch structure, which can learn the underlying feature information and guarantee the performance of each pixel feature for final classification. We also propose an edge supervised module to ensure that the segmentation boundary has a more accurate performance. A multiscale image fusion algorithm based on the Bayes model is proposed to improve the accuracy of the segmentation module. The performance of our model is evaluated on the ISPRS Vaihingen and Potsdam datasets. The results show that our method is superior to the most advanced image segmentation methods in terms of MIoU and pixel accuracy.

Fog degrades the quality of road images causing errors in stereo matching and road segmentation for Advanced Driver Assistance Systems (ADAS). Accident rates can be reduced if robust and efficient algorithms are applied for road image fog removal. Many studies have been conducted on this subject to date, but existing methods are not optimized for road images. Several daytime models cause local darkness, and blurring artifacts result in low quality haze-free images, and nighttime models are altogether inefficient. This study focuses on dehazing daytime and nighttime images by utilizing chromatic properties to remove haze from images. The proposed method treats fog as a specular pixels of dual consistency and physical properties, and the dehazing reflection model is suitable for parallel implementation to efficiently detect fog pixels. An edge-preserving low-pass filter (a fast-bilateral filter running 230x faster than an average CPU) is used to smooth the color components' original image maximum fraction to remove noise among fog pixels. The method significantly outperforms existing baselines in regards to both efficiency and dehazing.

Vision-based unstructured road following is a challenging task due to the nature of the scene. This paper describes a novel algorithm to improve the accuracy and robustness of vanishing point estimation with very low computational cost. The novelties of this paper are three aspects: 1) We use joint activities of four Gabor filters and confidence measure for speeding up the process of texture orientation estimation. 2) Misidentification chances and computational complexity of the algorithm are reduced by using a particle filter. It limits vanishing point search range and reduces the number of pixels to be voted. The algorithm combines the peakedness measure of vote accumulator space with the displacements of moving average of observations to regulate the distribution of vanishing point candidates. 3) Attributed to the design of a noise-insensitive observation model, the proposed system still has high detection accuracy even when less than 60 sparsely distributed vanishing point candidates are used for voting as the influence introduced by the stochastic measurement noise of vote function and the sparsity of the vanishing point candidates is reduced. The method has been implemented and tested over 20 000 video frames. Experimental results demonstrate that the algorithm achieves better performance than some state-of-the-art texture-based vanishing point detection methods in terms of detection accuracy and speed.