CN-122023412-A - Plastic mold surface defect image detection method and system

CN122023412ACN 122023412 ACN122023412 ACN 122023412ACN-122023412-A

Abstract

The invention discloses a method and a system for detecting surface defects of a plastic mold, which relate to the field of surface defect image detection of the plastic mold, are used for improving the identification capability of real defects, overcoming the detection difficulty caused by complex texture background in the prior art and improving the reliability and efficiency of detection, and comprise the steps of acquiring a mold surface image and multidimensional background information; the multi-dimensional background information comprises mold design drawing information, normal texture feature description and defect feature experience, analyzes a mold surface image, identifies a defect suspected area in the mold surface image, and judges the defect suspected area according to the defect suspected area and the multi-dimensional background information to obtain a defect judging result.

Inventors

XU WENXIANG
HE WEI
LIU XIUMIN
SHI ZHIBO

Assignees

惠州市广裕丰科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The plastic mold surface defect image detection method is characterized by comprising the following steps of: acquiring a mold surface image and multi-dimensional background information, wherein the multi-dimensional background information comprises mold design drawing information, normal texture feature description and defect feature experience; Analyzing the mold surface image, and identifying a defect suspected area in the mold surface image; And judging the suspected defect area according to the suspected defect area and the multidimensional background information to obtain a defect judging result.
2. The method of claim 1, wherein analyzing the mold surface image to identify a suspected region of defects in the mold surface image comprises: Determining average brightness and brightness variation ranges of a plurality of subareas in the mold surface image; And for any subarea in the plurality of subareas, if the average brightness of the subarea is larger than a preset brightness threshold value or the brightness variation range of the subarea is larger than a preset range threshold value, determining the subarea as a defect suspected area.
3. The method of claim 1, wherein analyzing the mold surface image to identify a suspected region of defects in the mold surface image comprises: and for any subarea in the plurality of subareas, if the chi-square distance between the LBP histogram of the subarea and the preset LBP histogram is larger than a first chi-square distance threshold value, determining the subarea as a defect suspected area.
4. The method for detecting a surface defect image of a plastic mold according to claim 1, wherein the step of judging the suspected defect area according to the suspected defect area and the multi-dimensional background information to obtain a defect judgment result comprises the steps of: Determining judgment weights and feature scores of a plurality of features according to the suspected defect area and the multidimensional background information, wherein the plurality of features comprise brightness features and texture features; according to the judging weight, carrying out weighted summation on the characteristic scores to obtain defect judging total scores; And under the condition that the defect judgment total score is larger than a score threshold value, determining the suspected defect area as a defect area.
5. The method for detecting a surface defect image of a plastic mold according to claim 4, further comprising: Calculating an aging degree index of the suspected defect region, wherein the aging degree index reflects the accumulated difference between the current texture state and the initial intact state of the region; Determining judgment weights of a plurality of features according to the suspected defect area and the multidimensional background information and according to the aging degree index.
6. The method for detecting surface defects of plastic molds according to claim 5, wherein the aging degree index of the suspected defect area satisfies the following relationship: Z t =α*L+(1-α)*Z t-1 ; Wherein Z t is the current aging degree index, Z t-1 is the aging degree index of the last detection period, alpha is a smoothing coefficient between 0 and 1, and L is the current chi-square distance.
7. The method for detecting surface defects of plastic molds according to claim 6, wherein said current chi-square distance satisfies the following relationship: L=0.5*sum((H1(i)-H2(i)) 2 /(H1(i)+H2(i))) wherein, H1 (i) is the count value of the LBP histogram of the current area in the ith bin, H2 (i) is the count value of the preset LBP histogram in the ith bin, and sum () is a sum function.
8. The method for detecting a surface defect image of a plastic mold according to claim 5, wherein determining the judgment weights of the plurality of features according to the defect suspected region and the multi-dimensional background information and according to the aging degree index comprises: Under the condition that the suspected defect area is in a predefined area, increasing initial brightness characteristic weight according to a first preset step length; and under the condition that the chi-square distance between the LBP histogram of the suspected defect area and the preset LBP histogram is larger than a second chi-square distance threshold, increasing the initial texture feature weight according to a second preset step length.
9. The method of claim 4, wherein determining feature scores for a plurality of features comprises: determining feature scores of brightness features according to the average brightness, the brightness change range and the first mapping relation of the suspected defect area, wherein the first mapping relation comprises different average brightness, different brightness change ranges and different brightness scores; And determining the feature scores of the brightness features according to the chi-square distance between the LBP histogram of the suspected defect area and the preset LBP histogram and a second mapping relation, wherein the second mapping relation comprises different chi-square distances and different texture feature scores.
10. A plastic mold surface defect image detection system, comprising: The image acquisition module is used for acquiring a mold surface image and multi-dimensional background information, wherein the multi-dimensional background information comprises mold design drawing information, normal texture feature description and defect feature experience; The abnormal region identification module is used for analyzing the mold surface image and identifying a defect suspected region in the mold surface image; And the defect judging module is used for judging the defect suspected area according to the defect suspected area and the multidimensional background information to obtain a defect judging result.

Description

Plastic mold surface defect image detection method and system Technical Field The invention relates to the field of plastic mold surface defect image detection, in particular to a plastic mold surface defect image detection method and system. Background In modern intelligent manufacturing plants, the detection of surface defects of plastic molds is critical to ensure product quality. However, conventional inspection methods often suffer from high false positive rates and high false negative rates when faced with molds having special surface textures, such as matte or natural texture-simulating molds. This is mainly because the complex texture and diffuse reflection characteristics of the mold itself interfere with image analysis, making it difficult for the system to effectively distinguish between true defects and inherent texture characteristics, thereby severely affecting production efficiency and product quality. Disclosure of Invention The application discloses a plastic mold surface defect image detection method and system, aiming at solving the technical problems that in the prior art, aiming at a mold with special surface textures, when the traditional online image detection system identifies surface defects, due to the interference of diffuse reflection characteristics and complex microscopic textures, the detection system has high false alarm rate and high false alarm rate, and the production efficiency and the product quality are seriously affected. In order to solve the technical problems, the invention provides a plastic mold surface defect image detection method, which comprises the steps of obtaining a mold surface image and multi-dimensional background information, wherein the multi-dimensional background information comprises mold design drawing information, normal texture feature description and defect feature experience; Analyzing the mold surface image, and identifying a defect suspected area in the mold surface image; And judging the suspected defect area according to the suspected defect area and the multidimensional background information to obtain a defect judgment result. According to the technical scheme, the defects on the surface of the die can be more accurately identified and judged by introducing multidimensional background information and combining with analysis of the surface image of the die, the problem that the false alarm rate and the false alarm rate are high when the special texture die is processed by the traditional method is effectively solved, and the reliability and the efficiency of detection are remarkably improved. Further, in some embodiments, analyzing the mold surface image identifies a suspected region of defect in the mold surface image, comprising: Determining average brightness and brightness variation range of a plurality of subareas in the mold surface image; And for any subarea in the plurality of subareas, if the average brightness of the subarea is larger than a preset brightness threshold value or the brightness variation range of the subarea is larger than a preset range threshold value, determining the subarea as a defect suspected area. According to the technical scheme, the potential abnormal region can be rapidly and effectively screened initially by analyzing the brightness and the brightness change range of the image subregion, a basis is provided for subsequent fine defect judgment, and the defect identification efficiency is improved. On the basis of the above, the application also provides that the analysis is carried out on the surface image of the mould, and the suspected defect area in the surface image of the mould is identified, which comprises the following steps: And for any subarea in the plurality of subareas, if the chi-square distance between the LBP histogram of the subarea and the preset LBP histogram is larger than a first chi-square distance threshold value, determining the subarea as a defect suspected area. According to the technical scheme, the LBP histogram and the chi-square distance are utilized to quantify the texture difference, so that the region inconsistent with the normal texture mode can be more accurately identified, and the method is particularly suitable for detecting the micro defect under the background with complex texture, thereby improving the accuracy of defect identification. Furthermore, the present application also proposes that the defect suspected area is judged according to the defect suspected area and the multidimensional background information to obtain a defect judgment result, which specifically comprises: Determining judgment weights and feature scores of a plurality of features according to the suspected defect area and the multidimensional background information, wherein the plurality of features comprise brightness features and texture features; according to the judgment weight, carrying out weighted summation on the feature scores to obtain defect judgment total scores; And under the co