Poor 2D edge detection

Problem

The results of machine vision algorithms that utilize 2D edge detection, e.g. 2D template matching and barcode scanning, are poor with Zivid 2 cameras.

Causes

The resolution of Zivid 2 cameras is 2.3 MP, which can be too low for 2D template matching, barcode scanning, and other machine vision algorithms that depend on good 2D edges. In addition, the demosaicing algorithm used with these cameras may leave image artifacts including RGB aliasing, erroneous interpolation direction and zipper effects.

Possible Solution

Replace your Zivid 2 camera with a Zivid 2+ camera. Zivid 2+ cameras have a 5MP image sensor, which means that more pixels are used to describe an image feature such as an edge. In addition, the demosaicing algorithm that Zivid 2+ cameras utilize is better than the one used with Zivid 2 cameras.

The figures below are from a 2D image taken with different Zivid cameras: Zivid 2+ M130 on the left side and Zivid 2 on the right side. In both cases, the 2D image is captured at a distance of 1.3 meters.

Barcodes

Barcode lines in 2D images taken with Zivid 2+ are sharp, while for the other cameras, they blend with the neighboring lines and have a strong zipper effect. The readability of the numbers and characters is also better with Zivid 2+.

Text

Text is more readable and lines are sharper and better preserved with Zivid 2+.

Patterns

Color patterns and lines are sharper and better preserved with Zivid 2+.