Image processing system for inspecting object distance and dimensions using a hand-held camera with a collimated laser

• US 10,489,928 B2
• Filed: 02/23/2018
• Issued: 11/26/2019
• Est. Priority Date: 02/23/2018
• Status: Active Grant

• Current Assignee: Librestream Technologies Incorporated
• Original Assignee: Librestream Technologies Incorporated
• Inventors: Shamilov, Radeon, Brewer, Scott
• Primary Examiner(s): Duley, Janese
• Application Number:
  US15/903,807
• Publication Number:
  US 20190266739A1
• Time in Patent Office: 641 Days
• CPC Class Codes: G06T 11/001 Texturing; Colouring; Generation of texture or colour inpainting G06T5/77
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/10028 Range image; Depth image; 3D point clouds
• CPC Class Codes: G06T 2207/20024 Filtering details
• CPC Class Codes: G06T 2207/20224 Image subtraction
• CPC Class Codes: G06T 5/40 using histogram techniques
• CPC Class Codes: G06T 7/521 from laser ranging, e.g. using interferometry; from the projection of structured light
• CPC Class Codes: G06T 7/55 from multiple images
• CPC Class Codes: G06T 7/62 of area, perimeter, diameter or volume
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: H04N 23/56 provided with illuminating means

Abstract

An image processing system for inspecting object distance and dimensions is disclosed. A calibrated hand-held camera with an integrated collimated laser is deployed in order to capture input images. The images are then processed in order to determine the object distance and related parameters such as displacements and surface dimensions.

Claims

1. A method for optically determining the distance to an object using a hand held camera, the method comprising the steps of:

acquiring using camera optics two sequential images of the target object wherein one of the images is a background image and the other of the images is a foreground image;

illuminating the target object in the foreground image with a collimated laser source that is laterally displaced from the camera optics to form a laser blob on the foreground image;

where the target object in the background image is not illuminated with the collimated laser source;

analyzing the background and foreground images in order to extract the laser blob coordinates in the image domain;

and calculating the distance to the object from the blob coordinates;

wherein said analyzing uses a background differencing step which generates a difference image between the background and foreground images;

and wherein in said analyzing both the foreground and background images are filtered using a convolved filter based on a scale independent version of the Bartlett window.

2. The method according to claim 1, wherein the Bartlett window is implemented by convolving a box filter of size K/2 with itself, where K relates to the Bartlett window of size K×

K and where K depends on the image resolution and it is set to be about the size of pixel diameter of the smallest laser blob the system is designed to detect, which in turn is implemented as separate convolution of two 1-dimensional moving average filters resulting in an O (M×

N) processing time where O refers to Big O notation and M and N represent the image size.

3. The method according to claim 1, wherein, before the background differencing step, the intensity of one color channel in the foreground image is enhanced relative to other color channels where the enhanced color channel corresponds to a dominant color of the collimated laser source.

4. The method according to claim 1, wherein a directional histogram spread is applied to the difference image produced by the background differencing step.

5. The method according to claim 4, wherein, in the histogram spread, individual pixels are enhanced in a manner that results with a top histogram bin having enough pixels to represent the smallest allowable laser blob.

6. The method according to claim 1, wherein the difference image produced by the background differencing step is used in a dynamic thresholding step using incremented threshold levels to produce a list of potential laser blobs.

7. The method according to claim 6, wherein the dynamic thresholding step uses sub-thresholds of at least one of pixel intensity, cluster size, cluster diameter, and cluster aspect ratio which are progressively adjusted to fit laser blobs that fall between the following two laser blob types:

Type A are laser blobs that correspond to one or more of the following;

Close target object;

Dark ambient light condition;

High intensity target object;

Type B are laser blobs that correspond to one or more of the following;

Far target object;

Bright ambient light condition;

Low intensity object.

8. The method according to claim 6, wherein a voting function is used to analyze the potential laser blobs based on at least one of their size, aspect ratio, and the original color in the foreground image, where the potential laser blob with the highest vote is selected and its center of mass is used to determine the distance from the camera to the illuminated object.

9. The method according to claim 8 wherein the voting function used is:

V=aS−

bA+cCG

Where;

a is the weight coefficient for the size of the laser blob;

S is the size of the laser blob in pixel count;

b is the weight coefficient for the aspect ratio;

A is the aspect ratio of the laser blob;

c is the weight coefficient for the overall color of the laser blob;

CG is the color grade computed for the laser blob.