Details and Options

• ImageSegmentationComponents is a high-level image segmentation that can either segment the whole image or segment specified objects and components in an image.

• The following specifications spec can be given:
• possegments a component specified by an {x,y} coordinate {pos₁,pos₂,…}attempts to segment the image into components corresponding to every pos_i {{pos₁₁,pos₁₂,…},{pos₂₁,…},…}specify each component with a list of positions Rectangle[…]specify each component with a bounding box {Rectangle[…],Rectangle[…],…}a list of bounding boxes, one for each component
• When multiple positions or rectangles are provided, a segment is computed for every marker. If different markers generate significantly overlapping components, the less dominant segment will be suppressed.
• The following options can be given:
• PerformanceGoal Automaticaspects of performance to try to optimize TargetDevice"CPU"the target device on which to perform training
• Possible settings for PerformanceGoal include:
• Automaticautomatic tradeoff between speed and quality "Quality"optimize for quality of final results "Speed"optimize for speed of getting results

Examples

Basic Examples  (1)

Segment an image:

Scope  (3)

Segment an image and colorize the label matrix:

Highlight segments on the image:

Generate a mask for a segment indicated by a list of positions:

Highlight the specified target points as well as the segmentation on the image:

Generate a mask for a segment inside a bounding box:

Highlight the bounding box as well as the segmentation on the image:

Options  (1)

PerformanceGoal  (1)

By default, a medium-speed model with moderate quality is used:

Aim for speed, but this typically results in lower quality:

Aim for high quality, but this typically leads to slower results:

Applications  (3)

Segment repeating components in an image:

Segment an x-ray or other image modalities:

Segment a region selected by points: