Case background

A cigarette factory found that the filling degree of cut tobacco was insufficient when packaging cigarettes, and there was also the reverse loading of cigarettes. Therefore, it was required to detect the situation of cigarettes after packaging.

Test items

• 01. Cigarette butts and cut tobacco sink, more than 1mm is considered ng.
• 02. Cigarette butts and cut tobacco are hollow, which is greater than 2 / 3 of the area of cigarette butts and is considered as ng.
• 03. Filter inverted
• 04. Less cigarette butts

Testing requirements

01. The speed is 600 packs / minute, the width of the mold box is 65mm, the height of the mold box is 30mm, the spacing of the mold box is 20mm, (65mm+20mm)*600/60s = 850mm/s
02. The total height of three smoke exhaust is about 25mm.

Solution

• 01. 3D line laser + depth learning algorithm based on semi supervised learning is adopted.
• 02. Do not need lighting, compatible with all products and lighting conditions.
• 03. Sr8060 sensor, the resolution of x-axis (line laser direction) is 0.012mm, and the resolution of y-axis (visual scanning direction) is 0. 65mm, z-axis resolution of 0.2um.
• 04. Highly stable and reliable intelligent identification of defects and statistics of defect types, which is convenient to trace product quality.