传感技术学报2025，Vol.38Issue(4)：623-633,11.DOI:10.3969/j.issn.1004-1699.2025.04.008
一种基于空间频率优化的各向异性轮廓检测方法
An Anisotropic Contour Detection Method Based on Optimization of Spatial Frequency
摘要
Abstract
Based on the modulation mechanism of spatial frequency to the difference in orientation,a biological visual contour detection method is proposed under spatial frequency modulation.Firstly,MCI contour detection model is introduced to perform primary contour extraction and multi-feature peripheral suppression calculations on the image.Then,on the basis of physiological mechanism that visual cells are sensitive to spatial frequency,a multi-layer spatial frequency detection model for contour detection is constructed.It improves the accuracy and robustness of frequency detection by introducing multi-directional spatial frequency perception based on neuron den-drites,spatial frequency filtering based on outlier elimination,and multi-level texture information integration based on the longitudinal arrangement of neurons.Finally,binary classification accuracy index is introduced to modulate towards selective suppression weights to achieve salient contour extraction of images.The P index and F index are adopted to evaluate the detection contours of RuG40 and BSDS300 databases,and the results are 0.63 and 0.76 respectively,37％and 16％higher than the original MCI method,indicating that the proposed model can make up for the lack of orientation difference in texture detection and effectively improve image quality.关键词
生物视觉/轮廓检测/朝向特征优化/空间频率调制Key words
biological vision/contour detection/orientation feature optimization/spatial frequency modulation分类
信息技术与安全科学引用本文复制引用
唐浩然,蔡哲飞,范影乐,武薇..一种基于空间频率优化的各向异性轮廓检测方法[J].传感技术学报,2025,38(4):623-633,11.基金项目
国家自然科学基金项目(61501154) （61501154）
