Semantic segmentation is an important method to interpret the road semantic environment. The convolution, pooling and deconvolution in semantic segmentation of deep learning result in blur and discontinuous segmentation boundary, missing and wrong segmentation of small objects. These influence the outcome of segmentation and reduce the accuracy of segmentation. To deal with the problems above, a new semantic segmentation method combined semantic boundary information was proposed. Firstly, a subnet of semantic boundary detection was built in the deep model of semantic segmentation, and the feature sharing layers in the network were used to transfer the semantic boundary information learned in the semantic boundary detection subnet to the semantic segmentation network. Then, a new cost function of the model was defined according to the tasks of semantic boundary detection and semantic segmentation. The model was able to accomplish two tasks simultaneously and improve the descriptive ability of object boundary and the quality of semantic segmentation. Finally, the method was verified on the Cityscapes dataset. The experimental results demonstrate that the accuracy of the method proposed is improved by 2.9% compared to SegNet and is improved by 1.3% compared to ENet. It can overcome the problems in semantic segmentation such as discontinous segmentation, blur boundary of object, missing and wrong segmentation of small objects and low accuracy of segmentation.

Intelligent vehicles need to use many characteristic parameters in unstructured road detection, which makes the feature fusion recognition difficult and computation complex, and the similarity of some road area and background may produce the mistake distinguishment and judgement of road identification. In order to solve the problems, an unstructured road detection method based on improved region growing with Principal Component Analysis-Support Vector Machine (PCA-SVM) rule was proposed. Firstly, the complex characteristic parameters such as color and texture of unstructured road were extracted, and then the PCA was used to reduce the dimension of the extracted characteristic information. The SVM trained with the primary characteristics reduced by PCA was used to be the classifier of the complex road cells. The priori knowledge such as the location of road, the initial cell and the characteristics of road boundary cells were used to improve the region growing method, and the classifier was used to decide the way of growing in cell growth for eliminating miscalculation area. The test results of actual roads show that, the proposed method has good adaptability and robustness, and can identify the unstructured road area effectively. The comparison results show that, compared with the traditional algorithm, the proposed method can shorten the calculation time by more than half through cutting characteristics from ten dimensions to three dimensions in ensuring the accuracy at the same time. The proposed method can also eliminate the 10% of miscalculation areas made by some similar areas of road and background for the traditional algorithm. The proposed method can provide a feasible way to shorten the recognition time and eliminate background interference in local path planning and navigation based on vision in the wild environment.