Automatic contour recognition method for tomographic images

Abstract

The invention discloses an automatic contour recognition method for a tomographic image. The method comprises the following steps: after an acquired tomographic image is de-noised, determining a boundary gray value T of a foreground and a background in a single image, and respectively correcting a foreground image gray value set and a background image gray value set by utilizing the boundary grayvalue T, a selected foreground image gray value and a selected background image gray value; and obtaining first foreground image feature data P2 and background image feature data P1, and inputting theforeground image feature data P2 and the background image feature data P1 through an onecut algorithm to obtain an identified second foreground image P3. According to the method, automatic identification of each image contour in multiple tissue slice images is realized, the method is particularly used for automatic processing of animal tissue slice image data with about thousands of layers, automatic contour information identification and extraction of the whole animal tissue slice tomographic image data are realized, and any identification parameters do not need to be modified manually.

Description

technical field

[0001] The invention belongs to image processing technology, and in particular relates to an image contour recognition technology. Background technique

[0002] Contour detection refers to the process of ignoring various noise interference inside the background and target in the digital image containing the target and the background, and adopting a certain algorithm to realize the process of extracting the target contour.

[0003] CN201710139018.3 discloses a target recognition method based on image contour features. In this method, the template image and the image of the object to be tested are preprocessed to generate a binary image. It is therefore necessary to create template image data. Not suitable for image processing of human or animal tissue slices.

[0004] CN109934813A discloses a method for identifying a target object. After identifying the first contour of the target object through a U-shaped full convolution neural network model, the method p...

Images