Biscuit breakage and internal defect detection method, system, storage medium and device

By combining multimodal image fusion and deep learning networks with visible light and infrared thermal imaging, high-sensitivity and high-precision detection of internal defects in biscuits is achieved. This solves the problems of single detection dimension and insensitivity to weak defects in traditional methods, and is suitable for achieving both quality and efficiency in high-speed production lines.

CN122289133APending Publication Date: 2026-06-26HANGZHOU HUICUI INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
HANGZHOU HUICUI INTELLIGENT TECH CO LTD
Filing Date
2026-02-27
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing technologies cannot effectively detect internal defects in biscuits, especially micro-cracks and internal structural anomalies. Furthermore, traditional methods are not sensitive to weak surface defects, making it difficult to balance detection speed and accuracy.

Method used

By employing multimodal image fusion technology, combining visible light and infrared thermal imaging, and using a deep learning model for defect detection, multi-channel semantic segmentation of biscuits is achieved using infrared physical feature extraction and deep learning networks to identify and quantify defect areas.

Benefits of technology

It enables comprehensive detection of internal defects in biscuits, improves detection sensitivity and accuracy, can distinguish defect types, provides feedback information for process parameter adjustment, and has robustness and adaptability, making it suitable for high-speed production lines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122289133A_ABST
    Figure CN122289133A_ABST
Patent Text Reader

Abstract

This invention discloses a method, system, storage medium, and device for detecting biscuit breakage and internal defects. The method includes: acquiring and preprocessing multimodal images to obtain registered images, wherein the registered images include a registered visible light image and a registered thermal image sequence; extracting infrared physical features from the registered thermal image sequence to obtain an infrared physical feature map; inputting the infrared physical feature map and the registered visible light image into a trained deep learning model to obtain a multi-channel semantic segmentation map, wherein each channel corresponds to a defect category; identifying defect regions based on the semantic segmentation map and performing region quantization to obtain a defect list; and generating a detection report for visualization by traversing the defect list in conjunction with preset quality standards. This invention achieves comprehensive and highly sensitive online automatic detection of biscuit damage, from macroscopic breakage to internal microscopic defects, through active thermal excitation and multimodal image fusion.
Need to check novelty before this filing date? Find Prior Art