A skin image information analysis method and analysis system

Abstract

This invention relates to the field of medical image processing technology and discloses a method and system for analyzing skin image information. For skin image analysis, the original image is first subjected to grayscale normalization and adaptive high-pass filtering to eliminate shooting differences. Gradients are then calculated and adaptively segmented using a median threshold to obtain a binary edge image. Local regions are cropped by centroid alignment and frequency domain transformation is performed to extract the phase spectrum matrix, retaining only structural information. This is then combined with normalized shape-invariant moments in the local coordinate system to form a joint vector fusing spatial and frequency domain features. Using a multi-sample reference center vector as a benchmark, and employing Euclidean difference metric and quantile adaptive thresholding, precise unsupervised determination of skin lesion regions is achieved.

Images