Affordable | Reliable | State of Art Vision Machine
The need of the hour for all manufacturing units is to maximize yield with minimum costs and resources and produce “zero” rejects, whether internal or external. Therefore, in any manufacturing process, it is crucial to detect the defects as early as possible in the process. Knowing about the problems early will allow immediate response to it and therefore will prevent the bad material to be passed on to the value added processes , reduce scrap& customer rejections and consequently increase yield.
A step in this direction is to make the surface defects “zero” which is imperative for meeting the strictest quality specifications of the customers by conducing inspections at every stage of production, detect & combat surface defects at the earliest.
This warrants for an inspection system that is reliable & efficient enough to detect all surface defects at every stage of manufacturing and accurate enough to classify them correctly, thereby enabling enhanced root cause analysis & elimination, faster quality grading and reduced need for manual inspection.
Tata Steel Automation Division offers you a vision product to tackle your challenging tasks of detecting surface defects automatically and to develop the kind of vision system that you need for your application.
SQUINS is an online system for detection of surface defects and tracking product performance at every stage.
The image capturing system offers options to operate in high sensitivity or low sensitivity mode, tall pixel mode and forward or reverse shift direction. When in high sensitivity mode, the camera uses both line scan sensors and its responsiveness increases accordingly. When in low sensitivity mode, the camera uses only the bottom sensor. Options for correction of false signals and non uniformity in the illumination profile; flat field correction; configurable region of interest and settings for line rate, exposure rate etc make the system adaptable and user friendly to a high degree.
Feature extraction module is for the purpose of extracting relevant defect features for the purpose of classification. Sixty seven such functions have been implemented for extracting various geometrical, textural, histogram and moment features.
Various ensembles of proven classifiers deliver high performance defect categorization. Different classifier recipes are incorporated ranging from standalone to classifier-pool-in systems to achieve outstanding performance.
The Training/Offline Classification Station is used to create the defect library. Experts view archived defect images and manually assign the “defect type” and severity level to them. All such “edited” images are compiled together to form a defect library which then goes as input to the Classifier for “Training the Classifier”.
This station shall have three display monitors. The monitors on the left and right side show a real-time display of images being acquired by the camera for the top & bottom side of the strip respectively and the monitor in the centre enables the operator to perform various tasks like configuring SQUINS, configuring camera parameters, analysing a defect image that has just been acquired, freezing an image that he has just seen on one of the two side monitors for analysis.