Covariance convolutional neural network-based low-contrast image saliency detection method

Abstract

The invention relates to a covariance convolutional neural network-based low-contrast image saliency detection method. The method comprises the steps of extracting low-level visual features of an image in a training set by taking a pixel as a unit; constructing a regional covariance based on a multidimensional eigenvector consisting of the low-level visual features; building a convolutional neuralnetwork model by taking a covariance matrix as a training sample; and based on local and global contrast principles, calculating image saliency. By performing test comparison on existing MSRA data set, SOD data set, CSSD data set, DUT-OMRON data set and PASCAL-S data set and an NI data set in the method, the method improves the robustness of conventional saliency detection, can effectively obtaina more accurate saliency map, can well extract a saliency target especially for the low-contrast image, and provides a very good solution for hot problems of night security monitoring, complex environment target locating and the like.

Description

technical field [0001] The invention relates to a low-contrast image saliency detection method based on a covariance convolutional neural network, which belongs to the technical field of image processing. Background technique [0002] In order to enable computers to process images as efficiently as humans, researchers have used the selective attention mechanism of the human visual system to propose visual saliency detection. Regions of interest, which can greatly improve the efficiency of computers in processing massive digital media information. Salient object detection in images provides a new way of thinking to solve problems in the field of computer vision and gradually occupies an important position. Salient detection can extract the main objects that human eyes pay attention to in image scenes. As a preprocessing module, reliable and fast saliency detection provides valuable reference information for applications such as segmentation and extraction of objects of inter...

AI Technical Summary

Problems solved by technology

Most of these existing visually salient object detection models are only suitable for visible light environments. However, in real life, some low-contrast scenes are often encountered, such as weather interference such as rain, snow, haze, or environments with poor lighting conditions at night. limit, which poses a great challenge to salient objec

Images