Method for automatically separating out the defect image from a thermogram sequence

• US 10,229,486 B2
• Filed: 05/30/2017
• Issued: 03/12/2019
• Est. Priority Date: 09/06/2016
• Status: Active Grant

• Current Assignee: University of Electronic Science and Technology of China
• Original Assignee: University of Electronic Science and Technology of China
• Inventors: Gao, Bin, Li, Xiaoqing, Tian, Guiyun
• Primary Examiner(s): Niu, Feng
• Assistant Examiner(s): Sun, Jiangeng
• Application Number:
  US15/608,320
• Publication Number:
  US 20180068432A1
• Time in Patent Office: 651 Days
• Field of Search: None
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 2207/20072 Graph-based image processing
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/0006 using a design-rule based approach
• CPC Class Codes: G06T 7/162 involving graph-based methods

Abstract

A method automatically separates out the defect image from a thermogram sequence based on the physical characteristics of the defect of a conductive material in electromagnetic field. Defect area radiates more heat than other area, when it is mapped to the histogram of the image to be separated, the defect area is located in the top-end of histogram, and the proportion of defect area is smaller to the background or other area. The method equally divides the histogram of the image to be separated into multi groups, and calculates the first derivative k_i of total pixel number S_i of group i, finds the maximum absolute value of k_i, i.e. |k_i|_max, where i is expressed as i_max; if i_max is the last group, W_n is regarded as threshold T, otherwise, W_imax+1 is selected as threshold T. The pixels that less than threshold T are removed to obtain the defect image in ECPT.

Claims

1. A method for automatically separating out a defect image from a thermogram sequence, the method comprising the following steps:

(1). pre-processing the thermogram sequence by Independent Component Analysis (ICA), including the following substeps;

obtaining a spatial and temporal heating response sequence (thermogram sequence) from a surface of a conductive material through an Eddy Current Pulsed Thermography (ECPT) testing system, and processing the thermogram sequence by conventional ICA, then, selecting an independent component with maximum kurtosis to reconstruct a defect image matrix according a size of an original thermogram, where the defect image matrix is an image to be separated, and;

(2). separating out the defect image based on a first derivative statistical characteristics of the image to be separated, including the following substeps;

(2.1). dividing a histogram of the image to be separated into n groups (n≥

2), where a magnitude of each group is calculated as follows;

2. The method for automatically separating out a defect image from a thermogram sequence of claim 1, wherein the group number n in step (2.1) is found through the following steps:

(2.1.1). initializing and coding a population, including;

randomly generating a number of individuals with equal probability, and taking them as an initial population {X₁, X₂, . . . , X_c, . . . , X_m}, where m is the size of the initial population, which corresponds to a number of the generated individuals, X_c is an individual of number c, a value of each individual is fallen in the scope that the group number is selected according to the histogram of the image to be separated;

(2.1.2). calculating a fitness of each individual, including;

taking each individual X_c in the initial population as a group number n, and separating the defect image R_C from the image to be separated according to step (2), then processing the defect image R_C into a binary image, by filling the pixels removed with 0, and by filling the pixels not removed with 1, and calculating a fitness H_C of the binary image as follows;

3. The method for automatically separating out the defect image from a thermogram sequence of claim 2, wherein the size in of the initial population is 10, the crossover probability is set as 0.4, the mutation probability is set as 0.3, and the number of generations of population is set as 7.