Standard Pattern Extraction Algorithm For Color Vision Test Map Based On Visual Saliency

After a long period of evolution,the human visual system has formed a selective visual attention mechanism that can quickly extract regions of interest from massive visual information.The saliency detection algorithm predicts and extracts the region of interest in the visual scene or image by simulating the visual attention mechanism.The color vision test map is drawn according to the principle of human eye color perception,which can well simulate the stimulation of the human eye in the natural environment,and the background is complex and the interference factors are more.Most of the known algorithms have problems such as insufficient biological basis and low detection accuracy,which is especially obvious when extracting the standard pattern in the color vision test map.In this paper,two novel saliency detection algorithms are proposed from different angles.In this paper,the calculation method of the normalization feature of the division is improved.By simulating the color double-antagonizing neurons in the human visual pathway,an adaptive channel weight color vision test map detection method based on the division normalization is proposed,which is based on the antagonistic neurons.The signal strength is calculated by adaptive weight calculation,and the final detection result is output after optimization.After studying the principles of visual psychology and cone-shaped cells and rod-shaped cells on the retina of human eyes,this paper proposes a method based on super-pixel segmentation combined with local features and global features to detect color vision test map.The pixel segmentation is preprocessed,and the local features and global features in a single channel are calculated by super pixel blocks instead of pixel points,and the channel weights are used for fusion to obtain the final detection result.We collected the commonly used color vision test map and scanned them,and produced excellent inspection map data sets and manual annotation maps.In this data set,the two algorithms proposed in this paper are compared with other algorithms,and the pattern extraction results are evaluated using visual comparison methods and objective evaluation indicators.Contrastive experiments show that compared with other classical algorithms,the two algorithms proposed in this paper can extract the standard patterns in the color vision inspection map more accurately.