Research On Image Texture Algorithm Based On LOTP

Texture is a basic property of the surface of the object and is widely used in image analysis.Texture analysis is the texture of the object image analysis,extraction of important surface grayscale information technology.Texture feature extraction and texture classification are hotspots in the field of image processing.In recent years,a variety of texture feature extraction methods have been endless.In this paper,a new local texture algorithm is proposed based on the original local texture descriptor,and the natural and biological images are extracted by Tamura's human visual features.Study on the Proper Classification of Natural Images and the Characteristics of Biological Images with Time.In order to extract the features of the image better,this paper proposes a local three valued model based on the local three-valued model,and describes the image with the characteristics of coarseness,contrast,directionality,regularity,line-likeness and roughness.The iterative decision tree based on the extracted features is used to classify the different types of natural images and human scar images of different ages.The results show that this method has higher correct rate and classification ability than other texture analysis methods,and it can accurately describe the image.In the process of studying the texture features of natural images,the Brodatz image library is selected,and the image is classified by using the iterative decision tree to establish the regression model.In the study of biological images,the second harmonic images of human scars were selected,and the regression model of normal human scars was used to predict the age of scar.Through the analysis of the model and the experimental results show that the proposed local descriptor and human visual characteristics can accurately describe the texture features of the image,with good texture description ability.