Class: Brief

Brief

new Brief()

Brief intends for "Binary Robust Independent Elementary Features".This method generates a binary string for each keypoint found by an extractor method.
Source:

Members

<static> N :number

The set of binary tests is defined by the nd (x,y)-location pairs uniquely chosen during the initialization. Values could vary between N = 128,256,512. N=128 yield good compromises between speed, storage efficiency, and recognition rate.
Type:
  • number
Source:

Methods

<static> getDescriptors(The, width, keypoints) → {Int32Array}

Generates a brinary string for each found keypoints extracted using an extractor method.
Parameters:
Name Type Description
The array grayscale pixels in a linear [p1,p2,...] array.
width number The image width.
keypoints array
Source:
Returns:
Returns an array where for each four sequence int values represent the descriptor binary string (128 bits) necessary to describe the corner, e.g. [0,0,0,0, 0,0,0,0, ...].
Type
Int32Array

<static> match(keypoints1, descriptors1, keypoints2, descriptors2) → {Int32Array}

Matches sets of features {mi} and {m′j} extracted from two images taken from similar, and often successive, viewpoints. A classical procedure runs as follows. For each point {mi} in the first image, search in a region of the second image around location {mi} for point {m′j}. The search is based on the similarity of the local image windows, also known as kernel windows, centered on the points, which strongly characterizes the points when the images are sufficiently close. Once each keypoint is described with its binary string, they need to be compared with the closest matching point. Distance metric is critical to the performance of in- trusion detection systems. Thus using binary strings reduces the size of the descriptor and provides an interesting data structure that is fast to operate whose similarity can be measured by the Hamming distance.
Parameters:
Name Type Description
keypoints1 array
descriptors1 array
keypoints2 array
descriptors2 array
Source:
Returns:
Returns an array where the index is the corner1 index coordinate, and the value is the corresponding match index of corner2, e.g. keypoints1=[x0,y0,x1,y1,...] and keypoints2=[x'0,y'0,x'1,y'1,...], if x0 matches x'1 and x1 matches x'0, the return array would be [3,0].
Type
Int32Array

<static> reciprocalMatch(keypoints1, descriptors1, keypoints2, descriptors2) → {Int32Array}

Removes matches outliers by testing matches on both directions.
Parameters:
Name Type Description
keypoints1 array
descriptors1 array
keypoints2 array
descriptors2 array
Source:
Returns:
Returns an array where the index is the corner1 index coordinate, and the value is the corresponding match index of corner2, e.g. keypoints1=[x0,y0,x1,y1,...] and keypoints2=[x'0,y'0,x'1,y'1,...], if x0 matches x'1 and x1 matches x'0, the return array would be [3,0].
Type
Int32Array