【Structure】 of image retrieval on Android

Retreive image from database

1. Use web service. – PHP

2. Save URLs of images in the database.

3. Retrieve URLs and download images directly from app in a separate thread.

4. Use JSON as the method of retrieval.

1) Server:

2) Client

  • take a query image – Use magnetic sensors on the phone to get the orientation of the camera.
    • Down-sample image to 1/16 of the original size
      1. mobile computing:
      2. cloud computing:
  • display result: how to display the returned image to the users.

Retrieving image from sd card

1. get all of the image that stored in sdcard

Existed Mobile Vision:

PhoneGuide[2] performs the computation on a mobile phone, instead of sending the images to a remote server. The system employs a neural network trained to recognize normalized color features and is used as a museum guide.

Roadside sign detection and inventory[3], uses GPS sensor, camera, and its algorithm is efficient enough to ensure good quality results in mobile settings.

AR object detection & recognition[4], use a modified version of the SIFT, CS, on a relatively small database of mobile phone imagery, capture image of urban environments and send to server for analysis

Camera tracking, extract SIFT feature from a video frame, then match it against features in database, use the correspondence to compute the camera pose.

Location telling[5], first build a “bootstrap database” of images (tagged with keywords) of landmarks, train a CBIR algorithm on it.  when a query image is matched against the bootstrap database, the associated keywords can be used to find more textually related images through a web search; CBIR algorithm is applied to the images returned from the web search to produce only those images that are visually relevant

the reason for image matching done directly on a mobile phone,

  1. the time to upload an image of the minimal resolution needed for successful matching exceeds the time to do the matching directly on the phone,  so reduce the system latency from 5-6 seconds to 2-3 seconds
  2. computational capabilities of mobile phones are expected to grow at a faster rate than the speed of wireless data networks
  3. distributing the computation among the users provides for better system scalability.


  • smaller images lead to faster processing: 640*480 to provide robust matching with sufficient speed
  • download speed is much faster than the upload speed
  • phone has the data cached locally
  • Bad data: In many cases, only a small percentage of image corresponds to the object of interest. The low resolution capture by low-end cameras with poor optics and noisy sensors.

    Using location information to limit irrelevant data, only considering data from nearby location cells, or loxels. incrementally updating the local database of features on the handset as the user changes location


[1] Hervé Jégou, Matthijs Douze, Cordelia Schmid “Hamming Embedding and Weak Geometry Consistency for Large Scale Image Search”, 10th European Conference on Computer Vision (ECCV ’08) 5302 (2008) 304–317

[2] E. Bruns and O. Bimber, “Adaptive Training of Video Sets for Image Recognition on Mobile Phones,” Journal of Personal and Ubiquitous Computing, 2008

[3] C. Seifert, L. Paletta, A. Jeitler, E. H¨odl, J.-P. Andreu, P. M. Luley, and A. Almer, “Visual Object Detection for Mobile Road Sign Inventory,” in Proc. of Mobile Human-Computer Interaction – Mobile HCI 2004, 6th International Symposium, 2004, pp. 491–495.

[4] G. Takacs, V. Chandrasekhar, B. Girod, and R. Grzeszczuk, “Feature Tracking for Mobile Augmented Reality Using Video Coder Motion Vectors,” in ISMAR ’07: Proceedings of the Sixth IEEE and ACM International Symposium on Mixed and Augmented Reality (ISMAR’07), 2007.

[5] T. Yeh, K. Tollmar, and T. Darrell, “Searching the Web with Mobile Images for Location Recognition,” in Proc. of Conference on Computer Vision and Pattern Recognition (CVPR), vol. 2. IEEE Computer Society, 2004, pp. 76–81.

List of CBIR


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s




Just another WordPress.com site

Jing's Blog

Just another WordPress.com site

Start from here......







Just another WordPress.com site

Where On Earth Is Waldo?

A Project By Melanie Coles

the Serious Computer Vision Blog

A blog about computer vision and serious stuff

Cauthy's Blog

paper review...

Cornell Computer Vision Seminar Blog

Blog for CS 7670 - Special Topics in Computer Vision


Life through nerd-colored glasses

Luciana Haill

Brainwaves Augmenting Consciousness



Dr Paul Tennent

and the university of nottingham

turn off the lights, please

A bunch of random, thinned and stateless thoughts around the Web

%d bloggers like this: