Halcon 10.0 模板匹配解决方案指南

5星 · 超过95%的资源需积分: 10 100 浏览量更新于2024-07-29 收藏 4.76MB PDF 举报

"HALCON 10.0 解决方案指南 II-B 匹配" 本手册是MVTec Software GmbH针对HALCON 10.0版本编写的解决方案指南，主要聚焦于匹配功能的使用，以帮助用户找到并定位对象。HALCON是一款强大的机器视觉软件，广泛应用于工业自动化、质量控制等领域。该指南旨在提供详尽的指导，使用户能够有效地利用匹配技术解决实际问题。匹配是计算机视觉中的一个关键步骤，它涉及到在图像中寻找与模板相似或相同的区域。在HALCON 10.0中，匹配功能经过优化，能够处理各种复杂情况，包括不同光照、角度变化、缩放和遮挡等。本指南分为多个部分，首先会介绍基本的匹配概念，包括模板匹配和模型匹配。模板匹配是通过与预定义的模板进行比较来寻找图像中的目标，而模型匹配则允许用户创建更复杂的模型，如基于形状或纹理的模型，以提高识别的准确性。接下来，手册将详细阐述如何设置和调整匹配参数，以适应不同的应用场景。这些参数可能包括相似度阈值、平移、旋转、缩放的允许范围等。用户还将学习如何使用不同的匹配算子，如灰度匹配、形状匹配和特征匹配，以及如何结合使用多种匹配方法以提高鲁棒性。此外，手册还会探讨如何处理匹配结果，包括如何评估匹配得分、如何选择最佳匹配候选，以及如何使用后处理技术如非极大值抑制来去除假阳性匹配。对于需要在三维空间中定位物体的情况，手册还将介绍相关的匹配策略和算法。在实际应用中，可能会遇到性能优化的问题。本指南会讲解如何利用HALCON的并行处理能力，以及如何调整内存使用，以在保持高精度的同时，提高匹配速度。最后，手册会给出一些实际案例，展示如何在具体项目中应用匹配技术，包括在质量检测、机器人引导和条码读取等场景中的应用实例。这些案例将帮助读者更好地理解匹配技术的实际效果，并提供实践指导。 "HALCON 10.0 解决方案指南 II-B 匹配"是一份详尽且实用的参考资料，对于任何使用HALCON进行图像处理和对象识别的工程师和技术人员来说，都是不可或缺的工具。通过深入学习和实践，用户可以充分利用HALCON的强大功能，解决各种复杂的视觉挑战。

B-16 Introduction

• ﬁgure 1.11 and ﬁgure 1.12 illustrate the transformations that can be handled by the individual

matching approaches,

• ﬁgure 1.13 lists the transformation parameters that are returned by the individual matching ap-

proaches, and

• ﬁgure 1.14 and ﬁgure 1.15 introduce typical changes of the appearance of objects and show which

characteristics of the appearance can be handled by the individual matching approaches.

2D object

component−based matching

gray−value−based matching

shape−based matching

application must be illumination−variant (rare−case)

object consists of components

between which angles and distances

may change, no scaling needed

occlusions, clutter, or color cannot be avoided

rather than clearly visible contours

object is characterized by specific texture

scaling is needed

images are highly defocused

object changes significantly

correlation−based matching

orthogonal view

one planar part of the object sufficient to

All three dimensions needed to

3D object

perspective deformable or

descriptor−based matching

3D matching

perspective view

distinguish the object from other objects

about the deformations needed

expected, rectification or information local deformable matching

none of the above stated restrictions or

local deformations of the object’s contour

Figure 1.10: First decision steps when selecting the matching approach.

1.5 Which Approach is Suitable in Which Situation? B-17

0000000000

1111111111

00000

11111

0000000000

1111111111

00000000

11111111

x,∆ ∆ y

RotationTranslation Similarity Measure

(Scale)

Anisotropic Scale

Sx = Sy

Figure 1.11: Geometric transformations of an object.

Matching Approach Translation Rotation Scaling Scaling

(2D) (2D) (uniform) (anisotropic)

gray-value-based X X - -

correlation-based X X - -

shape-based X X X X

component-based X X - -

local deformable X X X X

perspective deformable X X X X

descriptor-based X X X -

Figure 1.12: Transformations that can be handled by the matching approaches.

Matching Approach Position Angle Scale Scale Projective Pose

(2D) (2D) (uniform) (anisotropic) Transformation (3D)

Matrix

gray-value-based X X - - - -

correlation-based X X - - - -

shape-based X X X X - -

component-based X X - - - -

local deformable X - - - - -

perspective deformable - - - - X X

descriptor-based - - - - X X

Figure 1.13: Transformation parameters returned by the matching approaches.

Introduction

B-20 General Topics

2.1.1 Reduce the Reference Image to a Template Image

To create a model from a reference image, which is the common way for most matching approaches, the

reference image must be reduced to a template image that contains only those structures of the reference

image that are needed to derive the model.

For this, you select a region within the reference image that shows the part of the image that should serve

as the template or, respectively, from which the model should be derived. After selecting the region, the

domain of the reference image is reduced to an ROI using the operator reduce_domain. The resulting

image is our template image, which is input to one of the approach-speciﬁc operators that are provided

for the actual model generation.

Note that a region and therefore also the model can have an arbitrary shape (see the Quick Guide, sec-

tion 2.1.2.2 on page 17 for the exact deﬁnition of regions in HALCON). A region can be created in

different ways, as is described in the Solution Guide I, chapter 3 on page 19 and chapter 4 on page 31.

Summarized, a region can be selected, e.g., by the following means:

• A region can be speciﬁed by explicitly deﬁning its parameters. That is, HALCON offers mul-

tiple operators to create regions, ranging from standard shapes like rectangles (gen_rectangle2)

or ellipses (gen_ellipse) to free-form shapes (e.g., gen_region_polygon_filled). These

operators can be found in the HDevelop menu Operators . Regions . Creation. To use the

operators, you need the parameters of the shape you want to create, e.g., the position, size, and

orientation of a rectangle or the position and radius of a circle. If these parameters are not explic-

itly known, you can get them using draw operators, i.e., operators that let you draw a shape on the

displayed image and then return the shape parameters. These operators are available, e.g., in the

HDevelop menu Operators . Graphics . Drawing.

• A region can be speciﬁed by image processing using, e.g., a blob analysis. Then, the image is

segmented, e.g., using a threshold operator, and the obtained region is further processed to select

only those parts of it having speciﬁc features (commonly applied operators are, e.g., connection,

fill_up, and select_shape). In HDevelop, you can determine suitable features and values

using the dialog Visualization . Feature Inspection. More complex regions can be created

using set theory, i.e., by adding or subtracting standard regions using the operators union2 and

difference. That way, e.g., also ring-shaped ROIs can be created like is shown in section 2.1.2.2

on page 22.

Before creating the ROI, it is often suitable to optimize the reference image by a preprocessing. Note

that when using shape-based matching, you can use an HDevelop Assistant that guides you through the

matching process, which includes also the preprocessing of the reference image as well as the creation

of the ROI. How to use the Matching Assistant is described in the HDevelop User’s Guide, section 6.3

on page 221.

Note that also the gray values outside of the ROI have an inﬂuence on the model generation. In particular,

the inﬂuence is approximately 2

NumLevels

pixels, with n being the number of pyramid levels. Thus, the

gray values outside of the ROI selected in the reference image should be similar to those that will occur

in the search images.

剩余149页未读，继续阅读

laoniushitu

粉丝: 0
资源: 5

Halcon 10.0 模板匹配解决方案指南

使用Halcon进行1D高精度测量指南

全面了解HALCON机器视觉与HDevelop集成开发环境

Halcon-17.12-VS插件新工具：轻松查看HTuple和HObject变量

halcon-10.0-solution-guide-iii-a-1d-measuring

halcon-10.0

HALCON-10.0 libs.7z

halcon-10.0介绍与入门.ppt

C:\Program Files\MVTec\HALCON-10.0\include\cpp\HRegion.h(126,21): error C2059: 语法错误:“const”

halcon-6.1-shape-matching.pdf

halcon-13.0-hdevelop-users-guide

最新资源