Citation Context

...Ç 1 INTRODUCTION LOCAL feature matching has rapidly emerged to becomethe dominant paradigm for recognition and registration in computer vision. In traditional vision tasks such as panoramic stitching =-=[1]-=-, [2] and structure from motion [3], [4], it has largely replaced direct methods due to its speed, robustness, and the ability to work without initialization. It is also used in many recognition probl...

Citation Context

... [5], [6], [7], [8]. Feature-based methods are generally faster than global intensity alignment approaches, while they are more prone to local effects. A good survey on image registration is found in =-=[9]-=-. After computing the global transform chain, the second step is removing the annoying irregular perturbations. There are approaches which assume a camera motion model [3], [10], [11], [12], e.g., cam...

Citation Context

... up to 25% for websearch, and 40% for other key applications, simply from remapping application Table 1: Production Datacenter Applications applications description metric type content analyzer content and semantic analysis, used to take key words or text documents and cluster them by their semantic meanings [1] throughput latency-sensitive bigtable storage software for massive amount of data [9] average latency latency-sensitive websearch industry-strength internet search engine [5] queries per second latency-sensitive stitcher image processing and stitching, used for generating street views [43] N/A batch protobuf protocol buffer [2] N/A batch threads to cores. When co-locating threads from multiple applications, the optimal thread to core mappings changes. We also find that by leveraging knowledge of an application’s sharing characteristics, we can predict both how an application’s threads should be mapped when running alone as well as with another application. However, we conclude that using our online adaptive learning approach is a preferable approach for arriving at good thread to core mappings in the datacenter as it arrives at near optimal decisions and is agnostic to applicat...

Citation Context

...the oversensitivity to spatial structure. Another problem with template-based methods is that the objective function might not be smooth enough to reach the global optimum, as pointed out by Szeliski =-=[24]-=-. Often, the function is smoothed by blurring the image, for example using a Gaussian pyramid [1]. Recently, it has been proven that the Gaussian pyramid is not always the best choice for smoothing th...

Citation Context

...a set of globally registered images with limited overlapped region [22], [36]. For image mosaicing applications, global registration is performed based on a variety of predefined camera motion models =-=[33]-=-. For applications aiming to create special effect by engrafting image objects [1], [6], input images are initially registered by manual dragging or assuming static camera configuration. In all these ...

Citation Context

...e simple, fast, effective, and general. It is interesting to notice that to our best knowledge alg#1 may be one of the earliest color correction approaches developed for mosaic and panorama stitching =-=[27]-=-, while the direct predecessorofalg#6, alg#4 (global color transfer), is widely used as the baseline approach in color transfer research. After alg#1 and alg#6, alg#3 (tensor voting in joint image spa...

Citation Context

...h quality panorama generation is a well-known image stitching task. For a detailed overview of image stitching, the interested reader is referred to a comprehensive tutorial on this topic by Szeliski =-=[20]-=-. In most cases, the task of finding the geometrical relationship between individual images can be solved sufficiently well by determining image point correspondences, e.g., by using the well-known SI...