从入门到精通：OpenCV 2版计算机视觉教程

需积分: 10 180 浏览量更新于2024-07-23 收藏 11.12MB PDF 举报

"OpenCV（Open Source Computer Vision Library，简称OpenCV）是一个开源的计算机视觉库，其网址为<http://opencv.org>。该库主要用C和C++编写，支持多种操作系统，包括Linux、Windows、Mac OS X、iOS和Android，并提供了Python、Java、Ruby和Matlab等其他语言的接口，使得不同编程背景的开发者都能方便地利用其功能。 OpenCV的设计初衷在于追求计算效率，特别注重实时应用。它在算法到多核处理器和CPU指令等多个层次进行了优化，如支持SSE、MMX、AVX、NEON等技术，以提升性能。对于英特尔架构的进一步优化，用户可以选择Intel的Integrated Performance Primitives (IPP)库，这些库包含众多算法领域的低级别优化函数，OpenCV在运行时会自动调用最适合的指令。此外，OpenCV还提供GPU模块，包括对Nvidia GPU的CUDA加速版本和对通用GPU的OpenCL优化版本，以充分利用图形处理单元的并行计算能力。 OpenCV的目标是提供一个易于使用的计算机视觉基础设施，旨在帮助开发人员快速构建图像处理、物体检测、人脸识别、机器视觉等各种应用。其功能涵盖了基础图像处理（如图像读取、显示和转换）、特征检测和描述、图像分析（例如边缘检测、轮廓识别）、机器学习（如SVM分类）以及深度学习相关的工具，如卷积神经网络（CNN）的支持。通过这些功能，OpenCV成为了计算机视觉领域中最受欢迎的库之一，对于初学者和专业开发人员都是极其宝贵的资源。" OpenCV的学习路径通常分为几个阶段：首先，了解基本概念和库的结构，掌握C++或Python接口；其次，通过实践项目熟悉常用函数和模块，如图像读取、预处理、特征检测；接着深入理解算法背后的原理，如霍夫变换、SIFT/SURF特征提取；然后可以尝试更高级的主题，如目标跟踪、视频分析和深度学习技术；最后，随着技术进步，关注OpenCV的最新版本更新和社区扩展，持续学习和优化。通过这个系统的学习过程，新手能够逐步成长为OpenCV的高手。

• imgproc: The image processing module contains basic transformations on images,

including filters and similar convolutional operators.

• highgui: This HighGUI module contains user interface functions which can be

used to display images or take simple user input. It can be thought of as a very

light weight window UI toolkit.

• video: The video library contains the functions you need to read and write video

streams.

• calib3d: This module contains implementations of algorithms you will need to

calibrate single cameras as well as stereo or multi-camera arrays.

• features2d: The features2d module contains algorithms for detecting, describing,

and matching keypoint features.

• objdetect: This module contains algorithms for detecting specific objects, such as

faces or pedestrians. You can train the detectors to detect other objects as well.

• ml: The Machine Learning Library is actually an entire library in itself, and

contains a wide array of machine learning algorithms implemented in such as way

as to work with the natural data structures of OpenCV.

• flann: FLANN stands for “Fast Library for Approximate Nearest Neighbors”.

This library contains methods you will not likely use directly, but which are used

by other functions in other modules for doing nearest neighbor searches in large

data sets.

• gpu: The GPU library contains implementations of most of the rest of the library

functions optimized for operation on CUDA GPUs. There are also some functions

which are only implemented for GPU operation. Some of these provide excellent

results but require computational resources sufficiently high that implementation

on non-GPU hardware would provide little utility.

• photo: This is a relatively new module which contains tools useful for

computational photography.

• stitching: This entire module implements a sophisticated image stitching pipeline.

This is new functionality in the library but, like the ‘photo’ module is a place

where future growth is expected.

• nonfree: OpenCV contains some implementations of algorithms which are

patented or are otherwise burdened by some usage restrictions (e.g., the SIFT

algorithm). Those algorithms are segregated off to their own module, so that you

will know that you will need to do some kind of special work in order to use them

in a commercial product.

• contrib: This module contains new things that have yet to be blessed into the

whole of the library.

• legacy: This module contains old things that have yet to be banished from the

library altogether.

• ocl: The OCL module is a newer module, which could be considered analogous to

the GPU module, except that it relies on OpenCL, a Khronos standard for open

parallel computing. Though less featured than the GPU module at this time, the

OCL module aims to provide implementations which can run on any GPU or

other device supporting OpenCL. (This is in contrast to the GPU module which

explicitly makes use of the Nvidia CUDA toolkit and so will only work on Nvidia

GPU devices.)

#include "opencv2/calib3d/calib3d.hpp"

Calibration and stereo.

#include "opencv2/ml/ml.hpp"

Machine learning: clustering, pattern recognition.

#include "opencv2/highgui/highgui_c.h"

Old C image display, sliders, mouse interaction, I/O.

#include "opencv2/highgui/highgui.hpp"

New C++ image display, sliders, buttons, mouse, I/O.

#include "opencv2/contrib/contrib.hpp"

User-contributed code: flesh detection, fuzzy mean-shift tracking, spin images, self-similar features.

You may use the include file opencv.hpp to include any and every possible OpenCV function but, since it

includes everything, it will cause compile time to be slower. If you are only using, say, image processing

functions, compile time will be faster if you only include opencv2/imgproc/imgproc.hpp. These include

files are located on disk under the …/modules directory. For example, imgproc.hpp is located at

…/modules/imgproc/include/opencv2/imgproc/imgproc.hpp. Similarly, the sources for the functions

themselves are located under their corresponding src directory. For example, cv::Canny() in the

imgproc module is located in …/modules/improc/src/canny.cpp.

With the above include files, we can start our first C++ OpenCV program.

Legacy code such as the older blob tracking, hmm face detection, condensation tracker,

and eigen objects can be included using opencv2/legacy/legacy.hpp, which is located in

…/modules/legacy/include/opencv2/legacy/legacy.hpp.

First Program—Display a Picture

OpenCV provides utilities for reading from a wide array of image file types, as well as from video and

cameras. These utilities are part of a toolkit called HighGUI, which is included in the OpenCV package. On

the http://opencv.org site, you can go to the tutorial pages off of the documentation links at

http://docs.opencv.org/doc/tutorials/tutorials.html to see tutorials on various aspects of using OpenCV.

In the tutorial section, the “introduction to OpenCV” tutorial explains how to set up OpenCV for

various combinations of operating systems and development tools.

We will use an example from the “highgui module” to create a simple program that opens an image and

displays it on the screen (Example 2-1).

Example 2-1: A simple OpenCV program that loads an image from disk and displays it on the screen

#include <opencv2/opencv.hpp> //Include file for every supported OpenCV function

int main( int argc, char** argv ) {

cv::Mat img = cv::imread(argv[1],-1);

if( img.empty() ) return -1;

cv::namedWindow( "Example1", cv::WINDOW_AUTOSIZE );

cv::imshow( "Example1", img );

cv::waitKey( 0 );

cv::destroyWindow( "Example1" );

}

Note that OpenCV functions live within a namespace called cv. To call OpenCV functions, you must

explicitly tell the compiler that you are talking about the cv namespace by prepending cv:: to each function

call. To get out of this bookkeeping chore, we can employ the using namespace cv; directive as

shown in Example 2-2

. This tells the compiler to assume that functions might belong to that namespace.

Note also the difference in include files between Example 2-1 and Example 2-2; in the former, we used the

general include opencv.hpp, whereas in the latter, we used only the necessary include file to improve

compile time.

Example 2-2: Same as Example 2-1 but employing the “using namespace” directive. For faster compile, we

use only the needed header file, not the generic opencv.hpp.

#include "opencv2/highgui/highgui.hpp"

using namespace cv;

int main( int argc, char** argv ) {

Mat img = imread( argv[1], -1 );

if( img.empty() ) return -1;

namedWindow( "Example2", WINDOW_AUTOSIZE );

imshow( "Example2", img );

waitKey( 0 );

destroyWindow( "Example2" );

}

When compiled and run from the command line with a single argument, Example 2-1 loads an image into

memory and displays it on the screen. It then waits until the user presses a key, at which time it closes the

window and exits. Let’s go through the program line by line and take a moment to understand what each

command is doing.

cv::Mat img = cv::imread( argv[1], -1 );

This line loads the image.

The function cv::imread() is a high-level routine that determines the file

format to be loaded based on the file name; it also automatically allocates the memory needed for the image

data structure. Note that cv::imread() can read a wide variety of image formats, including BMP, DIB,

JPEG, JPE, PNG, PBM, PGM, PPM, SR, RAS, and TIFF. A cv::Mat structure is returned. This structure

is the OpenCV construct with which you will deal the most. OpenCV uses this structure to handle all kinds

of images: single-channel, multichannel, integer-valued, floating-point-valued, and so on. The line

immediately following

if( img.empty() ) return -1;

checks to see if an image was in fact read. Another high-level function, cv::namedWindow(), opens a

window on the screen that can contain and display an image.

cv::namedWindow( "Example2", cv::WINDOW_AUTOSIZE );

This function, provided by the HighGUI library, also assigns a name to the window (in this case,

"Example2"). Future HighGUI calls that interact with this window will refer to it by this name.

The second argument to cv::namedWindow() defines window properties. It may be set either to 0 (the

default value) or to cv::WINDOW_AUTOSIZE. In the former case, the size of the window will be the

same regardless of the image size, and the image will be scaled to fit within the window. In the latter case,

Of course, once you do this, you risk conflicting names with other potential namespaces. If the function foo()

exists, say, in the cv and std namespaces, you must specify which function you are talking about using either

cv::foo() or std::foo() as you intend. In this book, other than in our specific example of Example 2-2, we

will use the explicit form cv:: for objects in the OpenCV namespace, as this is generally considered to be better

programming style.

A proper program would check for the existence of argv[1] and, in its absence, deliver an instructional error

message for the user. We will abbreviate such necessities in this book and assume that the reader is cultured enough to

understand the importance of error-handling code.

剩余317页未读，继续阅读

greatwong

粉丝: 0
资源: 4

从入门到精通：OpenCV 2版计算机视觉教程

学习OpenCV中文版

学习OpenCV 3 中文版.rar

学习OpenCV(中文版-修改加书签)

推荐c++ 学习opencv

学习opencv3 英文

学习opencv需要多少时间

学习opencv pdf下载

学习opencv3 中文版

如何系统的学习opencv(c++)

脑电波学习是否需要学习opencv

最新资源