Channel logo segmentation method based on fully convolutional channel logo segmentation network

Abstract

A channel logo segmentation method based on a fully convolutional channel logo segmentation network includes the steps of constructing a multi-type fine-grained channel logo data set, the channel logoimage set containing a total number of 8400 images covering 42 categories; according to a channel logo region extraction method of pixel-by-pixel labeling, establishing a binary label image set corresponding to the channel logo image set and converting the binary label image set into an L-type single-channel gray image set; establishing a fully convolutional channel logo segmentation network of an end-to-end encoder-decoder structure, training the fully convolution label segmentation network on the channel logo data set, inputting a test image of any size in the channel logo image set to thetrained fully convolutional channel logo segmentation network, and generating a pixel-level segment result of the same size as the input image. The method of the invention helps the deep convolutionalneural network to realize powerful performance and solves the problem that the tiny target segmentation results are not precise enough. The network model can delineate a tiny target, thereby improving the spatial precision of the output, and is suitable for being applied to the tiny target segmentation method.

Description

technical field [0001] The invention relates to a station logo segmentation method. In particular, it relates to a logo segmentation method based on a fully convolutional logo segmentation network. Background technique [0002] Nowadays, with the emergence of various video software, massive videos are generated and disseminated in real time on network platforms, and the demand for applications such as classifying, recording and analyzing these videos is increasing day by day. Since the logo contains important information such as the source, orientation and category of the video, it is regarded as an important identification of the video. Algorithms to achieve video classification by segmenting logos emerged at the historic moment, which brought great challenges in terms of image segmentation technical requirements. [0003] Feature extraction is an important step in the process of logo segmentation, so finding more descriptive features has always been the goal pursued by r...

AI Technical Summary

Problems solved by technology

However, because the station logo field is generally composed of complex hollow structures and contains a lot of background noise, the existing image segmentation technology cannot completely separate the two

The color feature has strong separability, but for the translucent station logo, the color of the station logo usually changes due to the unevenness of the video picture or the influence of the background color, so the unstable color feature will bring difficulties to the station logo segmentation

The grid and principal component analysis feature treats all the content in the area equally, but due to the lack of full consideration of the possible hollow structure of the station logo, this feature is also greatly affected by the background

[0005] Due to the limited description ability of traditional artificial features, the expected segmentation effect cannot be achieved, but studies have shown that deep convolution features have stronger description capabilities than artificial features.

Images