The mine road is unstructured, the line shape is changeable, and the color of the road surface is similar to that of the mountains on both sides. In order to make up for the deficiency of the segmentation effect of the road detection model based on shape and color, a driving area segmentation method based on rut texture features is proposed. Initially, the collected images are preprocessed in the region of interest. Secondly, the gray level co-occurrence matrix (GLCM) is used to obtain the characteristic parameters of the rutting area, and four texture feature indexes are used as the input feature vectors of the genetic algorithm (GA) to obtain the optimal segmentation threshold. Finally, by filling the hole noise in the segmented image, a complete drivable area is obtained. The comparison test shows that the proposed segmentation method based on rut texture features can effectively overcome the problem that the road is similar to the background color and is difficult to segment. Compared with other similar segmentation methods, the segmentation effect of the road has better accuracy and robustness.
Utilizing machine vision to acquire road environment information is a crucial factor influencing the performance of advanced driver assistance systems, but under low-light conditions, obtaining high-quality road information directly through vision becomes challenging, necessitating image enhancement. Due to uneven illumination in low-light conditions, existing enhancement methods often lead to issues such as glare and blurred details in bright areas. In this paper, we propose an optimized Multi-Scale Retinex (MSR) image enhancement algorithm. Firstly, RGB images are converted to the YUV format, and the MSR algorithm is applied to enhance the Y channel. The local background brightness is then incorporated into the Just Noticeable Difference (JND) model to establish a non-linear relationship model between background brightness and adjustment factors. This enables adaptive adjustment of image enhancement intensity. Subsequently, a non-linear bilateral filtering function is applied to smooth the adjusted image, followed by Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance its contrast. Finally, the resulting image is fused with the original image in a 1:1 ratio. Experimental results indicate that, compared to the combination of MSR and histogram equalization, the proposed method achieves a 4.5% increase in standard deviation, a 7% increase in information entropy, and a 25% improvement in peak signal-to-noise ratio.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.