Object contour extraction method based on mask-RCNN

Abstract

The invention discloses an object contour extraction method based on mask-RCNN. The method comprises the steps: firstly obtaining a mask-RCNN model through trianing, inputting the RGB image under object contour extraction into mask-RCNN model for semantic segmentation, obtaining the corresponding binary mask image of the RGB image through processing of the mask-RCNN network, and inputting the RGBimage and the corresponding binary mask image into a contour refinement module. A contour feature description method is proposed to accurately represent the direction and angle information of the object contour, and the binary mask image contour obtained based on mask-RCNN is modified adaptively through the contour correction algorithm, and finally real-time accurate extraction of the image instance contour is realized under the condition of low image quality, low resolution, blurred object and low time and space complexity.

Description

technical field [0001] The invention belongs to the technical field of image object detection and segmentation of computer vision, and in particular relates to an object contour extraction method based on mask-RCNN. Background technique [0002] Forward-looking works on image detection and segmentation include the research on R-CNN (Regions with CNNfeatures) deep convolutional neural network proposed by Girshick et al. Kaiming et al. proposed the SPP-Net model to solve the above problems, and the processing speed is 30 to 170 times faster than R-CNN. In order to further reduce the time and space complexity of the instance segmentation algorithm, Girshick proposed the Fast-RCNN model to integrate feature extraction and classification into a classification framework, which improves the speed of training the model and the accuracy of target detection. Kaiming et al. added a branch network on the basis of Faster-RCNN to complete target pixel segmentation while realizing target ...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problem of inaccurate and incomplete segmentation of instance contour detection and incomplete segmentation due to factors such as image blur and low resolution in the existing image contour detection segmentation and extraction, and propose an object contour extraction based on mask-RCNN Method, first obtain a mask-RCNN instance segmentation model through training, propose a contour feature description method, accurately characterize the direction and angle information of the object contour, and use the contour correction algorithm to adaptively correct the instance contour obtained based on mask-RCNN , and finally achieve real-time accurate extraction of image instance contours when the image quality is not high, such as low resolution, blurred objects, and low time and space complexity.

Images