融合残差反卷积的图像分割算法研究OA

This paper proposes an image segmentation algorithm RDM-FCN that integrates residual deconvolution to address the issue of misclassification in FCN algorithm for processing complex scenes.In the encoder section,the VGG16 network is used to extract image features;In the decoder section,residual deconvolution modules are constructed and residual connections are introduced to enhance the transmission of cross layer features.By using the cross entropy loss function,the segmentation accuracy of the model is improved.The test results show that compared with the FCN algorithm,the accuracy of our algorithm has improved by 0.0347,the average intersection to union ratio has increased by 0.0215,and the average pixel accuracy has increased by 0.005.The experimental results show that the segmentation accuracy of the algorithm proposed in this paper is high,and it can effectively preserve the information of object edges and details.