Hands-On GPU-Accelerated Computer Vision with OpenCV and CUDA: Effective Techniques for Processing Complex Image Data in Real Time Using GPUs

Description

Discover how CUDA allows OpenCV to handle complex and rapidly growing image data processing in computer and machine vision by accessing the power of GPU Title Page......Page 2 Copyright and Credits......Page 3 Hands-On GPU-Accelerated Computer Vision with OpenCV and CUDA......Page 4 Packt Upsell......Page 5 Why subscribe?......Page 6 Packt.com......Page 7 Contributors......Page 8 About the author......Page 9 About the reviewer......Page 10 Packt is searching for authors like you......Page 11 Preface......Page 23 Who this book is for......Page 24 What this book covers......Page 25 To get the most out of this book......Page 27 Download the example code files......Page 28 Download the color images......Page 29 Code in Action......Page 30 Conventions used......Page 31 Get in touch......Page 32 Reviews......Page 33 Introducing CUDA and Getting Started with CUDA......Page 34 Technical requirements......Page 35 Introducing CUDA......Page 36 Parallel processing ......Page 37 Introducing GPU architecture and CUDA......Page 38 CUDA architecture......Page 40 CUDA applications......Page 42 CUDA development environment......Page 44 CUDA-supported GPU......Page 45 NVIDIA graphics card driver......Page 46 Standard C compiler......Page 47 CUDA development kit......Page 48 Installing the CUDA toolkit on all operating systems......Page 49 Windows......Page 50 Linux ......Page 52 Mac......Page 54 A basic program in CUDA C......Page 55 Steps for creating a CUDA C program on Windows ......Page 57  Steps for creating a CUDA C program on Ubuntu......Page 58 Summary......Page 59 Questions......Page 60 Parallel Programming using CUDA C......Page 61 Technical requirements......Page 62 CUDA program structure......Page 63 Two-variable addition program in CUDA C......Page 64 A kernel call......Page 67 Configuring kernel parameters......Page 69 CUDA API functions......Page 71 Passing parameters to CUDA functions......Page 73 Passing parameters by value......Page 74 Passing parameters by reference......Page 75 Executing threads on a device......Page 78 Accessing GPU device properties from CUDA programs......Page 80 General device properties......Page 81 Memory-related properties......Page 82 Thread-related properties......Page 83 Vector operations in CUDA ......Page 86 Two-vector addition program......Page 87 Comparing latency between the CPU and the GPU code ......Page 91 Elementwise squaring of vectors in CUDA......Page 92 Parallel communication patterns......Page 95 Map......Page 96 Gather......Page 97 Scatter......Page 98 Stencil......Page 99 Transpose ......Page 100 Summary......Page 101 Questions......Page 102 Threads, Synchronization, and Memory......Page 103 Technical requirements......Page 104 Threads......Page 105 Memory architecture......Page 109 Global memory......Page 111 Local memory and registers......Page 113 Cache memory......Page 114 Thread synchronization......Page 115 Shared memory......Page 116 Atomic operations......Page 119 Constant memory......Page 124 Texture memory......Page 127 Dot product and matrix multiplication example......Page 131 Dot product......Page 132 Matrix multiplication......Page 137 Summary......Page 142 Questions......Page 143 Advanced Concepts in CUDA......Page 144 Technical requirements......Page 145 Performance measurement of CUDA programs......Page 146 CUDA Events......Page 147 The Nvidia Visual Profiler......Page 150 Error handling in CUDA ......Page 154 Error handling from within the code......Page 155 Debugging tools......Page 157 Performance improvement of CUDA programs......Page 158 Using an optimum number of blocks and threads......Page 159 Maximizing arithmetic efficiency......Page 160 Using coalesced or strided memory access......Page 161 Avoiding thread divergence......Page 162 Using page-locked host memory......Page 163 CUDA streams......Page 164 Using multiple CUDA streams......Page 165 Acceleration of sorting algorithms using CUDA......Page 170 Enumeration or rank sort algorithms......Page 171 Image processing using CUDA......Page 174 Histogram calculation on the GPU using CUDA......Page 176 Summary......Page 182 Questions......Page 183 Getting Started with OpenCV with CUDA Support......Page 184 Technical requirements......Page 186 Introduction to image processing and computer vision......Page 187 Introduction to OpenCV......Page 189 Installation of OpenCV with CUDA support......Page 191 Installation of OpenCV on Windows......Page 192 Using pre-built binaries......Page 193 Building libraries from source......Page 194 Installation of OpenCV with CUDA support on Linux......Page 200 Working with images in OpenCV......Page 206 Image representation inside OpenCV......Page 207 Reading and displaying an image......Page 209 Reading and displaying a color image......Page 213 Creating images using OpenCV......Page 215 Drawing shapes on the blank image......Page 218 Drawing a line......Page 219 Drawing a rectangle......Page 220 Drawing a circle......Page 221 Drawing an ellipse......Page 222 Writing text on an image......Page 223 Saving an image to a file......Page 225 Working with videos in OpenCV......Page 226 Working with video stored on a computer......Page 227 Working with videos from a webcam......Page 230 Saving video to a disk......Page 232 Basic computer vision applications using the OpenCV CUDA module......Page 234 Introduction to the OpenCV CUDA module......Page 235 Arithmetic and logical operations on images......Page 236 Addition of two images......Page 237 Subtracting two images......Page 239 Image blending......Page 240 Image inversion......Page 241 Changing the color space of an image......Page 243 Image thresholding......Page 245 Performance comparison of OpenCV applications with and without CUDA support......Page 248 Summary......Page 252 Questions......Page 253 Basic Computer Vision Operations Using OpenCV and CUDA......Page 254 Technical requirements......Page 255 Accessing the individual pixel intensities of an image......Page 256 Histogram calculation and equalization in OpenCV......Page 258 Histogram equalization......Page 260 Grayscale images......Page 261 Color image......Page 263 Geometric transformation on images......Page 265 Image resizing......Page 266 Image translation and rotation......Page 268 Filtering operations on images......Page 270 Convolution operations on an image......Page 271 Low pass filtering on an image......Page 272 Averaging filters......Page 273 Gaussian filters......Page 275 Median filtering......Page 277 High-pass filtering on an image......Page 279 Sobel filters......Page 280 Scharr filters......Page 282 Laplacian filters......Page 284 Morphological operations on images......Page 286 Summary......Page 290 Questions......Page 291 Object Detection and Tracking Using OpenCV and CUDA......Page 292 Technical requirements......Page 294 Introduction to object detection and tracking......Page 295 Applications of object detection and tracking......Page 296 Challenges in object detection......Page 297 Object detection and tracking based on color......Page 298 Blue object detection and tracking......Page 299 Object detection and tracking based on shape......Page 302 Canny edge detection......Page 303 Straight line detection using Hough transform......Page 305 Circle detection ......Page 309 Key-point detectors and descriptors......Page 311 Features from Accelerated Segment Test (FAST) feature detector......Page 312 Oriented FAST and Rotated BRIEF (ORB) feature detection......Page 315 Speeded up robust feature detection and matching......Page 317 Object detection using Haar cascades......Page 322 Face detection using Haar cascades......Page 323 From video......Page 325 Eye detection using Haar cascades......Page 327 Object tracking using background subtraction......Page 329 Mixture of Gaussian (MoG) method......Page 330 GMG for background subtraction......Page 333 Summary......Page 337 Questions......Page 338 Introduction to the Jetson TX1 Development Board and Installing OpenCV on Jetson TX1......Page 339 Technical requirements......Page 340 Introduction to Jetson TX1......Page 341 Important features of the Jetson TX1......Page 344 Applications of Jetson TX1......Page 345 Installation of JetPack on Jetson TX1......Page 346 Basic requirements for installation......Page 347 Steps for installation......Page 348 Summary......Page 357 Questions......Page 358 Deploying Computer Vision Applications on Jetson TX1......Page 359 Technical requirements......Page 360 Device properties of Jetson TX1 GPU......Page 361 Basic CUDA program on Jetson TX1......Page 363 Image processing on Jetson TX1 ......Page 366 Compiling OpenCV with CUDA support (if necessary)......Page 367 Reading and displaying images......Page 369 Image addition......Page 371 Image thresholding......Page 373 Image filtering on Jetson TX1......Page 376 Interfacing cameras with Jetson TX1......Page 379 Reading and displaying video from onboard camera......Page 380 Advanced applications on Jetson TX1......Page 383 Face detection using Haar cascades......Page 384 Eye detection using Haar cascades......Page 387 Background subtraction using Mixture of Gaussian (MoG)......Page 389 Computer vision using Python and OpenCV on Jetson TX1......Page 393 Summary......Page 395 Questions......Page 396 Getting Started with PyCUDA......Page 397 Technical requirements......Page 398 Introduction to Python programming language......Page 399 Introduction to the PyCUDA module......Page 400 Installing PyCUDA on Windows......Page 401 Steps to check PyCUDA installation......Page 407 Installing PyCUDA on Ubuntu......Page 408 Steps to check the PyCUDA installation......Page 412 Summary......Page 413 Questions......Page 414 Working with PyCUDA......Page 415 Technical requirements......Page 416 Writing the first program in PyCUDA ......Page 417 A kernel call......Page 420 Accessing GPU device properties from PyCUDA program......Page 421 Thread and block execution in PyCUDA......Page 424 Basic programming concepts in PyCUDA ......Page 426 Adding two numbers in PyCUDA ......Page 427 Simplifying the addition program using driver class......Page 430 Measuring performance of PyCUDA programs using CUDA events......Page 432 CUDA events......Page 433 Measuring performance of PyCUDA using large array addition   ......Page 435 Complex programs in PyCUDA......Page 438 Element-wise squaring of a matrix in PyCUDA......Page 439 Simple kernel invocation with multidimensional threads......Page 440 Using inout with the kernel invocation......Page 443 Using gpuarray class......Page 445 Dot product using GPU array......Page 447 Matrix multiplication......Page 449 Advanced kernel functions in PyCUDA......Page 452 Element-wise kernel in PyCUDA......Page 453 Reduction kernel ......Page 455 Scan kernel ......Page 457 Summary......Page 459 Questions......Page 460 Basic Computer Vision Applications Using PyCUDA......Page 461 Technical requirements......Page 462 Histogram calculation in PyCUDA......Page 463 Using atomic operations......Page 464 Using shared memory......Page 467 Basic computer vision operations using PyCUDA......Page 470 Color space conversion in PyCUDA......Page 471 BGR to gray conversion on an image......Page 472 BGR to gray conversion on a webcam video......Page 474 Image addition in PyCUDA......Page 476 Image inversion in PyCUDA using gpuarray......Page 478 Summary......Page 480 Questions......Page 481 Assessments......Page 482 Chapter 1......Page 483 Chapter 2......Page 484 Chapter 3......Page 487 Chapter 4......Page 489 Chapter 5......Page 491 Chapter 6......Page 494 Chapter 7......Page 495 Chapter 8......Page 497 Chapter 9......Page 498 Chapter 10......Page 499 Chapter 11......Page 500 Chapter 12......Page 502 Other Books You May Enjoy......Page 504 Leave a review - let other readers know what you think......Page 506