Systems and methods for object detection

Patent metadata

Field	Value
Publication number	US-9508018-B2
Application number	US-201414551942-A
Country	US
Kind code	B2
Filing date	Nov 24, 2014
Priority date	Nov 24, 2014
Publication date	Nov 29, 2016
Grant date	Nov 29, 2016

Abstract

An object detection system and a method of detecting an object in an image are disclosed. In an embodiment, a method for detecting the object includes computing one or more feature planes of one or more types for each image pixel of the image. A plurality of cells is defined in the image, where each cell includes first through n th number of pixels, and starting locations of each cell in the image in horizontal and vertical directions are integral multiples of predefined horizontal and vertical step sizes, respectively. One or more feature plane summations of one or more types are computed for each cell. A feature vector is determined for an image portion of the image based on a set of feature plane summations, and the feature vector is compared with a corresponding object classifier to detect a presence of the corresponding object in the image portion of the image.

First claim

What is claimed is:

1. An object detection system for detecting an object in an image, the image comprising a plurality of image pixels, the object detection system comprising:
a memory to store instructions; and
a processor responsive to stored instructions to perform the steps of:
computing one or more feature planes of one or more types for each image pixel of the image;
defining a plurality of cells in the image, each cell comprising first through n th number of pixels, and starting locations of each cell in the image in horizontal and vertical directions being integral multiples of predefined horizontal and vertical step sizes, respectively;
computing one or more feature plane summations of one or more types for each cell of the plurality of cells, wherein a feature plane summation of a type for a cell is computed by summing corresponding features of the first through n th number of pixels of the cell, n being an integral number;
determining a feature vector for an image portion of the image, the feature vector comprising a set of feature plane summations of one or more cells defined in the image portion; and
comparing of the feature vector with a corresponding object classifier to detect a presence of the object in the image portion of the image.

2. The object detection system of claim 1 , wherein the one or more feature planes of one or more types comprise:
first, second and third feature planes for representing a color scheme of the image pixel;
a fourth feature plane for representing a gradient magnitude of the image pixel; and
fifth to m th Histogram of Gradient (HOG) feature planes for representing a gradient of the image pixel in corresponding fifth to m th HOG bins, m being an integral number.

3. The object detection system of claim 2 , wherein each HOG bin of the fifth to m th HOG bins is assigned a minimum binning angle and a maximum binning angle, and wherein a tangent of each of the minimum binning angle and the maximum binning angle is a rational number, and wherein the fifth to m th HOG feature planes of the image pixel are determined based on at least one of a shift operation, a subtraction operation and a comparison operation.

4. The object detection system of claim 3 , wherein the fifth to m th Histogram of Gradient (HOG) feature planes for an individual image pixel is computed by:
computing a horizontal gradient and a vertical gradient of the image pixel, and determining a corresponding HOG bin from among the fifth to m th HOG bins based on a ratio of the horizontal gradient and vertical gradient, wherein the corresponding HOG bin is determined by comparing the ratio with a first threshold value and a second threshold value associated with the corresponding HOG bin;
computing a gradient value of the image pixel; and
assigning the gradient value to a HOG feature plane corresponding to the determined HOG bin, and assigning zero values to remaining HOG feature planes of fifth to m th HOG feature planes.

5. The object detection system of claim 1 , wherein the processor is further responsive to stored instructions to perform the steps of:
computing a feature vector of a next image portion of the image portion of the image; and
comparing the feature vector of the next image portion with a corresponding object classifier, wherein the step of computing and comparing is performed in a sequential manner for each of remaining image portions of the image, thereby detecting the presence of the corresponding object in the image.

6. The object detection system of claim 5 , wherein a feature plane summation for a cell of predefined first horizontal and vertical cell sizes is computed based on corresponding feature plane summations of one or more cells of predefined second horizontal and vertical cell sizes, when the predefined first horizontal and vertical cell sizes are integral multiples of the predefined second horizontal and vertical cell sizes respectively.

7. The object detection system of claim 1 , wherein the processor is responsive to stored instructions to perform the steps of determining and comparing the feature vector by
performing in an iterative manner:
selecting a type of feature plane from the one or more types of feature planes;
computing a partial feature vector, the partial feature vector comprising one or more feature plane summations of the selected type, for one or more cells of the image portion;
comparing the partial feature vector with a corresponding partial classifier of the object; and
storing a result of a comparison in the memory;

collating one or more comparison results corresponding to each type of the feature plane of the one or more types of feature planes; and
detecting the presence of the object in the image based on the collated results.

8. The object detection system of claim 1 , wherein the processor is responsive to stored instructions to perform the step of scaling the image by a predefined scaling factor to obtain a scaled image.

9. A method of detecting an object in an image, the image comprising a plurality of image pixels, the method comprising:
computing one or more feature planes of one or more types for each image pixel of the image;
defining a plurality of cells in the image, each cell comprising first through n th number of pixels, and starting locations of each cell in the image in horizontal and vertical directions are integral multiples of predefined horizontal and vertical step sizes respectively;
computing one or more feature plane summations of one or more types for each cell of the plurality of cells, wherein a feature plane summation of a type for a cell is computed by summing corresponding feature planes of the first through nth number of pixels of the cell, n being an integral number;
determining a feature vector for an image portion of the image, the feature vector comprising a set of feature plane summations of one or more cells defined in the image portion; and
comparing the feature vector with a corresponding object classifier to detect a presence of the object in the image portion of image.

10. The method of claim 9 , wherein the computing the one or more feature planes of one or more types comprises:
computing first, second and third feature planes for representing a color scheme of the image pixel;
computing a fourth feature plane for representing an intensity of the image pixel; and
computing fifth to m th Histogram of Gradient (HOG) feature planes for representing a gradient of the image pixel in corresponding fifth to m th HOG bins, m being an integral number.

11. The method of claim 10 , wherein each HOG bin of the fifth to m th HOG bins is assigned a minimum binning angle and a maximum binning angle, and wherein a tangent of each of the minimum binning angle and the maximum binning angle is a rational number, and wherein the fifth to m th HOG feature planes of the image pixel are determined based on at least one of shift operation, subtraction operation and comparison operation.

12. The method of claim 9 , wherein the computing the fifth to m th Histogram of Gradient (HOG) feature planes for an individual image pixel comprises:
computing a horizontal gradient and a vertical gradient of the image pixel, and determining a corresponding HOG bin from among the fifth to m th HOG bins based on a ratio of the horizontal gradient and vertical gradient, wherein the corresponding HOG bin is determined by comparing the ratio with a first threshold value and a second threshold value associated with the corresponding HOG bin;
computing a gradient value of the image pixel; and
assigning the gradient value to a HOG feature plane corresponding to the determi

Assignees

• Texas Instruments Inc

Inventors

• Desappan Kumar Arrakutti
• Mathew Manu
• Swami Pramod Kumar

Classifications

• G06V10/507 Primary

  Summing image-intensity values; Histogram projection analysis · CPC title

• G06V10/56

  relating to colour · CPC title

• G06V10/758

  Involving statistics of pixels or of feature values, e.g. histogram matching · CPC title

• G06K9/6212

  Physics · mapped topic

• G06K9/4647 Primary

  Physics · mapped topic