MEGATRADE Corporation

Dedicated to Designing what you mean by gVisual Inspectionh

Algorithms

Brings Results that Properly Reflect Customersf Demands

We are continuously developing additional image processing algorithms to meet customersf demands. Through a decade of specializing in the development of visual inspection software, MEGATRADE has now reached its height of expertise especially in the field of gHigh-speed comparison inspection of printed circuit boards (PCBs) with complicated circuitry.h Let us now introduce a number of major inspection algorithms:

Self-segmentation of Target Area

Target area image of specimen is divided automatically into several parts such as pattern area, space area etc.  Different criteria can be applied to different parts.  Strict inspection is carried out, focusing on important part area.

Self-segmentation of Target Area
Brightness Thresholding

An algorithm that detects candidate defects out of brightness distribution of 256 shades of gray.  Two thresholds, one at bright side and the other dark side, can be set. 

Brightness Thresholding
Cluster Thresholding

Algorithm for discriminate candidate defects by the size of dot group.  It judges not simply by the number of NG dots in a block, but also by another criteria that those dots form cluster or not.

Cluster Thresholding
Contour Examination

Contour part is first highlighted particularly, and then is subjected to inspection by this algorithm.  This algorithm has a feature that it goes along with wide contour variation and detects small contour variation as defect.

Contour Examination
Pixel-by-Pixel Check

Every pixel of specimen image are compared to predetermined thresholds for luminance and number of contiguous out-of-threshold pixels.  This process requires a number of reference pieces for threshold setting, but has the advantage that it can save time for master data generation.

Pixel-by-Pixel Check
Automatic Luminance Calibration

Average luminance of reference piece and that of specimen to be tested are automatically adjusted to coincide.  Subtle tint variation between one lot and another can be ignored by this calibration.

Automatic Luminance Calibration
Segmented Position Adjustment

Three points on a specimen are designated, with which specimen position is to be adjusted.  In case that target area is divided into parts, positions of every part are to be adjusted independently. 

Segmented Position Adjustment

Technology