MATLAB VIDEO AND IMAGE PROCESSING BLOCKSET 3 Guide de l'utilisateur télécharger pdf (Page 28)

Video and Image Processing Blockset™ Release Notes

Performance Improvements

The Vide o and Image Processing B locks et contains three performance

enhancements for R2007b.

Label Block Has More Efficient Labeling Algorithm

The old block implemen ta tion was based on a flood-fill algo rith m, while the

new algorithm is union-find based. As a result, the new labeling algorithm

uses less mem ory and is a more cache efficient.

Compa tibility Consi derations. The new labeling algorithm might label the

objects in the image with different label n um bers.

Morphological Blocks Experience Speed Increase

The speed of simulation and the speed at w hich your generated code runs will

increase, if you meet all of the following conditio ns:

• You are using any of the blocks in the M orphological Operations library

except the Label block.

• You are working with i ntensity images.

• The structuring element you are using has four or more contiguous

elements that form a horizontal line, a v ertical line, or a shape that the

block can decompose into a se t of horizontal lines, vertical lines, or both.

For more information about structuring elements, see the

strel function

reference page in the Image Processing Toolbox documen tatio n.

Row-Major Data F ormat Supported by Chroma Resampling,

Deinterlacing, and Insert Text blocks

The Chroma Resampling, Deinterlacin g, and Insert Tex t blocks give you the

option to process data that is stored in r ow -major format when you se lect the

Input image is transposed (data order is row m ajor) check box. For

more information, see the Chroma Resampling, D einterlacing, and Insert

Text block reference pages.

1 2 ... 23 24 25 26 27 28 29 30 31 32 33 ... 58 59

Pas de commentaire

MATLAB VIDEO AND IMAGE PROCESSING BLOCKSET 3 Guide de l'utilisateur Page 28