Double Lane Line Edge Detection Method Based on Constraint Conditions Hough Transform

Image processing is a key link in the process of automobile intelligence. Image pretreatment methods are studied, including gray processing, binarization and extraction of regions of interest. Three different methods of gray image are compared, and the maximum interclass variance method is used for binarization processing of images. When the image is too bright or too dark, the recognition is very inaccurate. Then the edge detection and straight line acquisition were studied to identify the lane lines, and the Canny edge detection was studied, including Gaussian filtering smoothing, Sobel operator, non-maximum suppression and double threshold edge extraction, and the Canny algorithm was improved and optimized to study the Hough transform. Hough transform is used to extract the straight lines to recognize the lane lines.

Lane detection plays a key role in building intelligent traffic system. How to improve the accuracy of lane recognition and the ability of curve detection has always been the focus of research. Aiming at the detection of yellow and white lane lines and curves in complex environments, this paper introduces a new lane line detection method based on machine vision, which combines the yellow lane line processed in HSV space with the white lane line processed in grayscale space. Through canny edge detection, inverse perspective transformation and sliding window polynomial fitting method to achieve real-time lane detection. The experiment shows that the algorithm can accurately detect the curve and the lane line under the light changing circumstances, and can calculate the vehicle's deviation distance according to the lane line detection results. At the same time, the algorithm shows good reliability and robustness under multiple working conditions.

Lane line detection plays an important role in vehicle autonomous driving, however non lane line markings and vehicle shadows seriously influence lane line detection results. In this paper, we propose a new lane line segmentation and detection algorithm based on inverse perspective mapping (IPM). Firstly, by using the restriction of the consistent lane line width after IPM, we segment the lane line and project the segmentation results back to the original coordinate system; then detect the right and left two lane lines by the gray scale weighted Hough transform method. Experimental results show that the proposed algorithm has strong robustness to non lane line markings and vehicle shadows, and the lane line can be accurately detected under the complex pavement conditions.

A Robust Lane Detection Method Based on Vanishing Point Estimation

Lane line detection is the key to construct intelligent transportation system. Improving the capability of lane line recognition and curve detection accuracy has been an important part of the research. For the problems of slow speed and low precision of the traditional lane line recognition of yellow lane and the difficulty of Hough transform recognition of curve lane line, a lane line detection method using machine vision is proposed. This method combines white lane line extracted from gray space with yellow lane line extracted from HSV space. Real-time lane detection can be realized by inverse perspective transformation, canny edge detection and sliding window polynomial fitting, and the deviation distance of vehicles can be calculated according to the detection results to carry out deviation warning. The test results indicate that this method is able to accurately identify the straight and curved lane lines of multi-cycle urban roads, and shows good reliability and robustness.

It has always been an important low-level operation to extract edges from images in the fields of computer vision and image procession, in which straight line extraction is typical and representative. Because most man-made spatial objects, e.g. buildings, roads, etc. often take on near straight-line boundaries, extracting straight lines is often the first step to extract these targets. Straight lines can then be looked as the elementary units for other higher level image interpretations. In this paper, a straight line extraction method combining edge detection and depth-first searching on the vector line layer is proposed and applied to extract runways of airports. The steps include: 1) edges are found with the Canny operator and vectorirzed. The reason to use the Canny operator is because it is designed to be an optimal edge detector, which gives very good results on detecting step or slop like edges. It takes as input a grey scale image, and produces as output an image showing the positions of tracked intensity discontinuities. After this operation, we then vectorize the edge points to be a vector layer with edge tracing.2) With the vector-formatted edge lines, the straight line searching can then be carried out. In order to complete this, topology between arcs should be cleaned and rebuilt, which includes the deletion of repetitive, one-node arcs, and splitting on the intersections, etc. 3) Straight lines are detected with the depth-first searching strategy. With the rebuilt topology, we can easily obtain the begin, end nodes of every line. If the distances of its all vertices to the line connecting the begin, end nodes of an arc are less than some pre-defined threshold, it could be looked as a 'straight line' and extracted. Besides, we are certainly only interested in the straight lines with lengths larger than certain threshold, thus a minimum length threshold should be specified to delete these very short lines. In the searching of straight lines, some arcs should be grouped as a single straight line; some un-straight lines should be split to extract its straight parts. The suitable straight lines are outputted to a vector layer after being reselected and re-grouped, with distinguishing short, long isolating, long not isolating straight lines. With all these steps, we can get the initial straight vector line layer. 4) To these lines with small interspaces but locate on a single straight line, we use a simple but effective connecting step to 'fill' the gaps. Starting from the vector layer and with the operations of broken line connecting and parallel line detection, the main airport runway can be well extracted, which helps us to locate and recognize airports from high spatial remotely sensed imagery.

Lane line detection is an important part of intelligent driving. However, the road pictures taken in haze weather have low clarity, and the loss of detailed information brings great difficulties for lane line detection. The classical defogging algorithm cannot meet the accuracy requirements of lane line detection, so this paper proposes a foggy lane line detection algorithm for improving the dark channel and Canny operator. First, the foggy image is divided into the area of interest, and then use the quadruple tree multilevel search to obtain the accurate value of atmospheric light. The restored image is image-enhanced to improve the effect of fog removal; The images are then preprocessed for edge detection using the modified Canny operator, and we use a hybrid filter instead of Gaussian filter. The maximum inter-class variance (OSTU) algorithm obtains the adaptive optimal threshold and improves the experimental performance of the algorithm. Finally, the polar angle constraint uses the Hough transformation to accurately fit the lane lines. Experiments show that the improved defogging algorithm obtains higher image contrast, clearer image, and the edge detects more preserved lane line information, and improves the recognition accuracy of vehicle lane lines.

The robustness and stability of lane detection is vital for advanced driver assistance vehicle technology and even autonomous driving technology. To meet the challenges of real-time lane detection in complex traffic scenes, a simple but robust multilane detection method is proposed in this paper. The proposed method breaks down the lane detection task into two stages, that is, lane line detection algorithm based on instance segmentation and lane modeling algorithm based on adaptive perspective transform. Firstly, the lane line detection algorithm based on instance segmentation is decomposed into two tasks, and a multitask network based on MobileNet is designed. This algorithm includes two parts: lane line semantic segmentation branch and lane line Id embedding branch. The lane line semantic segmentation branch is mainly used to obtain the segmentation results of lane pixels and reconstruct the lane line binary image. The lane line Id embedding branch mainly determines which pixels belong to the same lane line, thereby classifying different lane lines into different categories and then clustering these different categories. Secondly, the adaptive perspective transformation model is adopted. In this model, the motion information is used to accurately convert the original image into a bird’s-eye view image, and then the least-squares second-order polynomial fitting is performed on the lane line pixels. Finally, experiments on the CULane dataset show that the proposed method achieved similar or better performance compared with several state-of-the-art methods, the F1 score of the proposed method in the normal test set and most challenge test sets is better than other algorithms, which verifies the effectiveness of the proposed method, and then the field experiments results show that the proposed method has good practical application value in various complex traffic scenes.

Lane detection is a key technology in automatic driving systems. Aiming at the problems of low accuracy and susceptible to interference in the traditional detection algorithm as for yellow lane lines, and poor recognition of curved lane lines by the traditional Hough transform. In this paper, a lane line detection method based on machine vision is proposed. Firstly, the white lane lines are identified in the gray space, the yellow lane lines are identified in the HSV space and the yellow lane lines and white lane lines are then merged. The interference is reduced by dividing the region of interest and Gaussian filtering. Then, the edge of the lane line is identified by edge detection and the image is transformed into an aerial view by inverse perspective transformation. Finally, the sliding window polynomial fitting method is adopted to fit the lane lines. At the same time, a lane departure warning method based on calculating curve curvature and center deviation is proposed.

Many cars now have functions to assist the driver in driving, such as keeping the lane. This function can keep the vehicle in a proper position between lanes. This function can ensure the safety of the vehicle. The traditional lane detection method relies on image processing and has poor robustness. At the same time, due to the need for post-processing technology, it will generate a lot of time-consuming calculations, which is not conducive to use in changing road scenes. Based on this, this paper proposes a two-stage lane line detection method based on deep neural network, which divides lane detection into two stages, lane edge segmentation and lane line classification. In the first stage, a semantic segmentation network based on a lightweight network performs pixelated segmentation on the edge of lane lines. In the second stage, a long-term storage network based on GRU is used to detect lane lines based on lane edges. Practice has proved that the lane line detection method proposed in this paper is very reliable in various driving scenarios. At the same time, due to the use of various lightweight networks, the network model in this paper can be deployed on vehicle systems with powerful real-time performance.

Nowadays, autonomous driving is becoming more and more popular. Lane line detection is very important for trajectory planning and decision making in autonomous driving. Traditional lane detection methods rely on highly defined, manual feature extraction and heuristic methods, which usually require post-processing technology. More and more recently, the approach is modeling with deep learning. The lane line scheme based on segmentation usually requires large model and complex convolution structure design, and it cannot perceive the lane line geometric features. Similar to the heat map scheme, the detection of the key points of the lane line actually belongs to the same scheme as the segmentation in a certain angle, but it only reduces part of the amount of computation. The current methods all ignore the data imbalance between the lane line categories that the near lane line occupies most of the position of the picture, resulting in far lane line samples are far less than the near samples. In this paper, a novel detection scheme for key points of lane lines is proposed. The key points of lane lines are linearly sampled at different intervals on the longitudinal axis of images to solve the problem of data imbalance between lane lines. Then the sampled anchor points are fixed, and the model only needs to predict the abscissa of each lane line at the anchor points. At the same time, the geometric constraint loss function of the lane line is put forward to ensure the correct lane line shape. The method presented in this paper achieves 50 FPS on embedded devices, it achieved SOTA on the Culane and Tusimple datasets.

Lane line images have the essential attribute of large-scale variation and complex scene information, and the similarity between adjacent lane lines is high, which can easily cause classification errors. And remote lane lines are difficult to recognize due to visual angle changes in width. To address this issue, this paper proposes an effective lane detection framework, which is a hybrid feature fusion network that enhances multiple spatial features and distinguishes key features throughout the entire lane line segment. It enhances and fuses lane line features at multiscale to enhance the feature representation of lane line images, especially at the far end. Firstly, in order to enhance the correlation of multiscale lane features, a multi-head self attention is used to construct a multi-space attention enhancement module for feature enhancement in multispace. Secondly, a spatial separable convolutional branch is designed for the jumping layer structure connecting multiscale lane line features. While retaining feature information of different scales, important lane areas in multiscale feature information are emphasized through the allocation of spatial attention weights. Finally, considering that lane lines are elongated areas in the image, and the background information in the image is much more abundant than lane line information, the flexibility of traditional pooling operations in capturing widely existing anisotropic contexts in actual environments is limited. Therefore, before embedding feature output branches, strip pooling is introduced to refine the representation of lane line information and optimize model performance. The experimental results show that the accuracy on the TuSimple dataset reaches 96.84%, and the F1 score on the CULane dataset reaches 75.9%.

Lane detection technology is of great significance in intelligent transportation. Standard Hough transform is easy to generate false peak value and to cause error detection of line when extracting line. In this paper, the wavelet algorithm is combined to extract the area of interest in lane line recognition, and then the wavelet reconstruction and Canny algorithm are used to obtain the image edge, and the lane line is detected by the Hough transform algorithm based on linear parameter space reduction. This algorithm can improve the correct detection rate of lane lines, avoid the interaction between lane lines and reduce the misdetection of both sides of the lane. The improved algorithm is more conducive to the real-time control of intelligent vehicle assistance and automatic driving in time.

Lane departure warning systems and lane line detection methods based on image processing and semantic segmentation: A review

Lane detection is an important component of vehicle-assisted driving systems. Many lane line algorithms have been proposed, but lane line detection is still a challenging task in environments with road text or vehicle interference. To address these issues, this article proposes a robust lane line detection method under structured roads. First, we propose a method based on the slope and length of a straight line and the distance of the line from the center line to extract regions of interest reducing interference from other invalid regions. Afterward, we use a new fusion method to fuse the extracted color features with the gradient features, enhancing the lane features while greatly reducing the interference in the pixels. Next, we cut the extracted pixel feature maps into equal parts into several sub-maps and perform pixel statistics for each sub-map, replacing sub-maps with abnormal pixel values with adjacent sub-maps, which can effectively reduce noise. Finally, we use density-based spatial clustering of applications with noise (DBSCAN) to cluster the pixels and fit the lane lines with least squares. The experimental results show that the method can effectively detect lane lines and has good robustness.