News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
Image Recognition Feature Extraction
Image Recognition Feature Extraction
Image Recognition Feature Extraction
HOG features:
The Histogram of Oriented Gradient (HOG) feature is a feature descriptor used to stop object detection in computer vision and image processing. It forms features by calculating and counting the gradient direction histograms of the partial regions of the image. Hog feature separation SVM classifier has been widely used in image recognition, especially in pedestrian detection. The demand hints that the HOG+SVM approach to stop pedestrian detection was proposed by French researcher Dalal at the 2005 CVPR. Of course, there are always many pedestrian detection algorithms proposed from time to time, but they are mainly HOG+SVM thoughts.
(1) Main ideas:
In an image, the appearance and shape of a partial purpose can be well described by the distribution of the density of gradients or edges. (Substantial: Gradient statistics, while gradients exist mainly in the center of the edge).
(2) The detailed completion method is:
First divide the image into small connected areas. We call it the cell unit. A gradient or edge direction histogram of each pixel in the cell unit is then acquired. Finally, the histograms can be combined to form a feature descriptor.
(3) Progress performance:
These parts of the histogram are stopped within the larger range of the image (we called it interval or block) to be contrast-normalized. The approach used is to: first calculate each histogram in this interval (block Density in the cell, and then normalizes each cell in the interval based on this density. After this normalization, you can achieve better results with changes in lighting and shadows.
(4) Advantages:
HOG has many advantages compared to other features. First, since HOG is operated on some grid cells of an image, it maintains good invariance to both geometric and optical deformations of the image, and these two deformations will only appear in a larger spatial category. Second, under the conditions of coarse airspace sampling, precise direction sampling, and strong partial optical normalization, only pedestrians can generally maintain an upright posture, allowing pedestrians to have some subtle movements, these subtle movements. Can be neglected without affecting the detection effect. The HOG feature is therefore particularly suitable for human detection in images.
2. The completion of the HOG feature extraction algorithm:
About the process:
HOG feature extraction method is to an image (the purpose or scan window you want to detect):
1) Grayscale (see the image as an x, y, z (grayscale) 3D image);
2) The Gamma correction method is used to normalize the color space of the input image (normalization); the purpose is to adjust the degree of illumination of the image, reduce the influence of shadow and light changes in the image part, and at the same time suppress noise interference;
3) Calculate the gradient (including size and direction) of each pixel of the image; mainly to capture the contour information while further weakening the illumination interference.
4) divide the image into small red cells (eg 6*6 pixels/cell);
5) Statistics each cell's gradient histogram (the number of different gradients), can constitute a descriptor for each cell;
6) Each cell is composed of a block (for example, 3*3 cells/block). A feature descriptor of all cells in a block is connected in series to obtain a HOG feature descriptor of the block.
7) A concatenation of the HOG feature descriptors of all the blocks in the image image can obtain the HOG feature descriptor of the image (the object you want to detect). This is the ultimate feature vector available for classification.
HOG features:
The Histogram of Oriented Gradient (HOG) feature is a feature descriptor used to stop object detection in computer vision and image processing. It forms features by calculating and counting the gradient direction histograms of the partial regions of the image. Hog feature separation SVM classifier has been widely used in image recognition, especially in pedestrian detection. The demand hints that the HOG+SVM approach to stop pedestrian detection was proposed by French researcher Dalal at the 2005 CVPR. Of course, there are always many pedestrian detection algorithms proposed from time to time, but they are mainly HOG+SVM thoughts.
(1) Main ideas:
In an image, the appearance and shape of a partial purpose can be well described by the distribution of the density of gradients or edges. (Substantial: Gradient statistics, while gradients exist mainly in the center of the edge).
(2) The detailed completion method is:
First divide the image into small connected areas. We call it the cell unit. A gradient or edge direction histogram of each pixel in the cell unit is then acquired. Finally, the histograms can be combined to form a feature descriptor.
(3) Progress performance:
These parts of the histogram are stopped within the larger range of the image (we called it interval or block) to be contrast-normalized. The approach used is to: first calculate each histogram in this interval (block Density in the cell, and then normalizes each cell in the interval based on this density. After this normalization, you can achieve better results with changes in lighting and shadows.
(4) Advantages:
HOG has many advantages compared to other features. First, since HOG is operated on some grid cells of an image, it maintains good invariance to both geometric and optical deformations of the image, and these two deformations will only appear in a larger spatial category. Second, under the conditions of coarse airspace sampling, precise direction sampling, and strong partial optical normalization, only pedestrians can generally maintain an upright posture, allowing pedestrians to have some subtle movements, these subtle movements. Can be neglected without affecting the detection effect. The HOG feature is therefore particularly suitable for human detection in images.
2. The completion of the HOG feature extraction algorithm:
About the process:
HOG feature extraction method is to an image (the purpose or scan window you want to detect):
1) Grayscale (see the image as an x, y, z (grayscale) 3D image);
2) The Gamma correction method is used to normalize the color space of the input image (normalization); the purpose is to adjust the degree of illumination of the image, reduce the influence of shadow and light changes in the image part, and at the same time suppress noise interference;
3) Calculate the gradient (including size and direction) of each pixel of the image; mainly to capture the contour information while further weakening the illumination interference.
4) divide the image into small red cells (eg 6*6 pixels/cell);
5) Statistics each cell's gradient histogram (the number of different gradients), can constitute a descriptor for each cell;
6) Each cell is composed of a block (for example, 3*3 cells/block). A feature descriptor of all cells in a block is connected in series to obtain a HOG feature descriptor of the block.
7) A concatenation of the HOG feature descriptors of all the blocks in the image image can obtain the HOG feature descriptor of the image (the object you want to detect). This is the ultimate feature vector available for classification.