Simulation research on optimal extraction of architectural image features based on image sequence

Abstract

The optimized extraction of ancient architectural image features in Jiangnan can more accurately describe the geometric information of architectural image. To extract the image features, we need to classify the image pixel features, calculate the response function value of each image pixel, and complete the feature optimization extraction of the image. The traditional method reconstructs the image and uses morphological operation to detect the edge of the building image, but ignores the calculation of the response function value of each pixel. An optimal feature extraction method of image in Jiangnan ancient architecture based on SUSAN corner detection is proposed. The Gaussian mixture model is used to represent the color distribution corresponding to each pixel of Jiangnan ancient architecture image, describe the characteristics of each pixel of the image, classify the characteristics of each image pixel, define the corner response function by calculating the gradient value of image pixel gray change in Jiangnan ancient architecture, and calculate the corner response function value of each image pixel, The local maximum image pixel that can obtain this value is obtained as the image feature of Jiangnan ancient architecture. The experimental results show that the proposed method can effectively improve the accuracy of feature optimization extraction in the image of ancient buildings in the south of the Yangtze River, and has good robustness.