A method and system for image-based multi-target segmentation and recognition

Abstract

The invention discloses an image-based multi-target segmentation recognition method and system, belonging to the field of image recognition. The method includes: extracting a single-channel image of an image to be recognized and performing binarization processing to obtain the outermost contour of each object respectively and the innermost contour; approximate the outermost contour and the innermost contour of each object into a polyline, and use the angle formed by two adjacent polylines to screen out the candidate points for segmentation; find and segment the candidate points by the principle of the closest distance within the area Corresponding target segmentation candidate points to form segmentation point pairs; connect each segmentation point pair in the binarized image to separate each object, and extract the target outermost contour of each object; if the inner area of ​​the target outermost contour is smaller than the preset If the area threshold is set, the corresponding segmentation point pairs are deleted to obtain the segmented object images; the image features of each object image are extracted through the deep neural network model, and the classifier determines the category to which each object belongs. The present invention has better segmentation effect.

Description

technical field [0001] The invention belongs to the field of image recognition, and more particularly, relates to an image-based multi-target segmentation and recognition method and system. Background technique [0002] One of the problems in the existing automatic image recognition systems for food retail platforms such as bread, vegetables, and fruits is that it is difficult to separate the images of each item so that it can be recognized by the recognition system after each item is in contact with each other. At present, many schemes use deep learning for image segmentation, but deep learning requires a large amount of labeled data and runs slowly. For food retail platforms such as bread, vegetables, and fruits, it is necessary to adopt a simple and fast image segmentation method based on edge detection. [0003] Currently, there are methods for identifying in the following ways: 1. Extracting bending points from the contour of the bread image to connect the bending poin...

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects or improvement needs of the prior art, the present invention proposes an image-based multi-target segmentation and recognition method and system, thereby solving the technical problem that existing methods cannot adapt to target recognition in complex situations

Images