Image Pre-Processing

Image Pre-Processing is enabled by default.

 

AutoVISION’s Image Pre-Processing algorithms are available for the following tools:

 

 

Image Pre-Processing makes it easier for the software to recognize and read characters in an image, detect features in an image, and measure width, height, and angle within an image. The Pre-Processing routine runs before the tool runs.

 

The Image Pre-Processing editor shown here is visible in the Decode Tool, OCR Tool, Count Tool (Blob Count), and Measure Tool’s parameters areas.

 

Separate Objects (Open)

Separate Objects (Open) uses morphological pre-processing to perform an open operation. This operation is useful when you want to separate objects from each other, such as multiple connected OCR characters that are connected and are misidentified as a single character. Once this option is selected, the editor is updated to provide parameters specific to the open operation.

 

 

Separate Objects (Open) Example:

 

Connect Objects (Close)

Connect Objects (Close) uses morphological pre-processing to perform a close operation. This operation is useful when you want to connect objects to each other, such as multiple dot-matrix-printed OCR characters whose dots are too far apart, making it difficult to segment the characters. Once this option is selected, the editor is updated to provide parameters specific to the close operation.

 

 

Connect Objects (Close) Example:

 

Shrink Objects (Erode)

Shrink Objects (Erode) uses morphological pre-processing to perform an erode operation. This decreases the size of objects. This operation is similar to Separate Objects in that it is typically used to separate objects that are touching. The difference is that the Separate Objects operation performs an erosion followed by a dilation, so that objects are separated but then returned to their original size. The Shrink Objects option only erodes objects. The editor options for this operation are the same as for Separate Objects.

 

 

Shrink Objects (Erode) Example:

 

Grow Objects (Dilate)

Grow Objects (Dilate) uses morphological pre-processing to perform a dilate operation. This increases the size of objects. This operation is similar to Connect Objects option in that it is typically used to connect objects. The difference is that the Connect Objects operation performs a dilation followed by an erosion, so that objects are connected but then returned to their original size. The Grow Objects option only dilates objects. The editor options for this operation are the same as for Connect Objects.

 

 

Grow Objects (Dilate) Example:

 

Gradient Image (Sobel)

Gradient Image (Sobel) performs a Sobel operation on the image, which highlights areas where there is contrast. The resulting image is dark where there is no contrast and light where there is contrast. This option is often used when trying to detect features in an area where illumination is uneven or inconsistent. Use this option with the Count Tool for applications in which trying to set a grayscale threshold is nearly impossible, and in which you need to detect objects based on their degree of contrast with the background.

 

 

Gradient Image (Sobel) Example:

 

Smooth Noise (Mean Low Pass Filter)

This option smoothes out background “noise” by softening image contrast via a weighted neighborhood average operation. It is typically used on specular, metallic backgrounds.

 

 

Smooth Noise (Mean Low Pass Filter) Example: