Visual Search

It is a technical to search objects in visual environment, like search your friend in a crowd of people, search for food in a supermarket. My visual search have used a eye movements as means to measure to the degree of attention given to stimuli.

But vast research suggest that eye movements move independently of attention and therefore is not a reliable method to examine the role of attention. Much of the previous literature on visual search uses reaction time in order to measure the time taken to detect the target amongst its distractors.

==Search type:
Features search. The objects could be describe by the features. And the target could be quickly figured out by checking the value of the features. For example, from the red circle in a set of black circle. This search could be very efficient. It can be done in parallel.

Conjunction search. When the target and the distractors have more similarities in more than one single visual property such as size, color, orientation, and shape.

Image matching. Visual search via image matching can be used for a large variety of applications. It should have one or more search libraries containing images to be matched, associated with metadata such as descriptions, classifications, and links to further information. The advantages of this method


==Theory,

One way to visual search via image of regions can be accomplished is using a mathematical technique called vector quantization. It is like partitioning the image into kernels, similar to the technique of tokenizing the document into works or sentences. Then, the image is stored as indexes of these kernels.