- Date:17 Apr 2020
- Views:19
- Downloads:0
- Pages:8
- Size:220.34 KB

Transcription:

QIONG et al TIE DYE TECHNIQUE AND PATTERN FEATURES 181. identify and separate the shadow of moving vehicle samples were soaked at 100 for 120 min and. Yang Hongyin et al 8 come up with an image indexing separately put into 5 different dyeing levels Table 1 at. method on the basis of HSV space color edge 60 C and 100 regain Table 1 shows the required. histogram amount gram of reactive dyes and auxiliaries for. So on comparing the models of RGB and YCrCb 100g fabric reactive black KN B and reactive red. HSV color space is found better to describe the color B 2BF are reactive dyes and sodium sulphate. HSV color space has two important features anhydrous SSA and soda ash SA are two. viz V component is not related to color information auxiliaries contents in water Auxiliaries in all 5 vats. H component and S component are closely related to also contain 0 5g L detergent 0 5g L glacial acetic. the style how the human eyes observe the colors acid 1g L anti staining agent and 2g L softener Each. These features make the HSV color space suitable for sample was prepared with three rotation speeds. digital image processing So it is thought to analyse namely 5 circles min 15 circles min and 30 circles. tie dye image within HSV color space min After dyeing the fabrics were washed and dried. Texture is a key component of human visual Finally they were scanned by high definition scanner. perception It describes the homogeneity of image with resolution of 300dpi The images 27cm 42cm. surface and the spatial distribution of different were collected and stored with RGB mode Fig 1. elements not depending on the color and brightness. information It reflects the global and local structure 2 2 Methods. of images and is widely used for image retrieval The processing of tie dye image includes two steps. Texture methods can be categorized as statistical viz image preprocessing and extraction of pattern. geometrical structural model based and signal features. processing features There are many texture feature Preprocessing includes the procedure of image. extraction methods including wavelet based texture space conversion smooth filtering and image. feature Tamura feature and feature gray level segmentation The pattern information includes the. co occurrence matrix9 Tamura10 features have average value of HSV tri component of the valid. six dimensions including coarseness contrast tie dye area the proportion of tie dye white area of. directionality line likeness regularity and roughness the image and Tamura texture. which take into account human perception of texture The scanned image was saved in RGB color. The last three components can be derived from the model which produces 24 digit color space through. first three So the first three have been used for the sensitivity quantity of sensitive components of. texture analysis in this paper R G and B at each pixel The conversion formula. Based on the digital image processing technology from the RGB model to. this paper extracts the average value of HSV, tri component of the valid tie dye area the proportion HSV model is as follows. of tie dye white area and Tamura texture as pattern 0 if max min. features A set of variance experiments are designed 60 0 if max R and G B. to validate the strong relationship between the G B. 60 360 if max R and G B, tie dye technique and the characteristic quantity of H max min. color and texture Then a mathematic model is built B R. 60 120 if max G, up to forecast the tie dye technique the tie dye max min. concentration rotation speed with the characteristic 60 240 if max B. quantity of color and texture,0 if max 0,2 Materials and Methods 1 else. 2 1 Materials,The experiments were carried out under the indoor.

temperature of 25 We selected 18 2tex 140g m2 RGB color matrix has been normalized within. cotton knitted fabric and cut into size of 0 1 with linear function max is the maximum value. 60cm 40cm These fabrics were then pinched and and min is the minimum value among R G and B. tied tightly with space of about 1cm apart Then the The scope of H value after conversion ranges from. 182 INDIAN J FIBRE TEXT RES JUNE 2016,Fig 1 Experimental samples. 0 to 360 and S value and V value are in the range,0 1 For the subsequent treatment H is normalized. to the range of 0 1 with linear function,2 2 1 Image Processing. 2 2 1 1 Noise Reduction,The tie dye images interfere within the complex. environment so it is necessary to filter the Fig 2 Filtering effect diagram of two examples. The median filter is the nonlinear digital filtering. technique It has the advantages of easy calculation. and high feasibility The filter adopts a sliding,window with odd points In each matrix of HSV.

the median of every elemental value replaces the,gray value of the central point If the number. of elements is odd the value in the middle is,Fig 3 Segmentation image of two examples. the median If the number of elements is even,the average value of two values in the middle. is the median This paper adopts 3 3 median filter where t is the threshold t 0 L 1 w0 the. background proportion u0 the background average,to eliminate the noise the filtered images become. value w1 the foreground proportion u1 the foreground. smoother Fig 2, average value and u the average value of the whole.

2 2 1 2 Image Segmentation image The value t which makes the expression. The study adopts Otsu algorithm to segregate the formula above maximum is the best threshold value. tie dye images into dyeing area and white area of image segmentation Fig 3. Assume L is the largest gray level in gray level, image Otsu11 method describes the following 2 2 2 Feature Extraction. criterion for selecting the optimum threshold t 2 2 2 1 Color Feature. Color features of the image include the average, max t w 0 t u 0 t u 2 w 1 t u 1 t u 2 2 value of HSV tri component of valid tie dye area and. QIONG et al TIE DYE TECHNIQUE AND PATTERN FEATURES 183. proportion of white area Calculation formula of HSV E k x y max Ek h x y Ek v x y. average value is defined as 8, 1 Step 4 Take the average of Sbest over the picture to. H i j be the coarseness measure Fcrs as shown below. S m S i j Fcrs i 1 j 1 Sbest i j m n, Vm V i j i j W 3 where m and n are the effective size of the image. where W is the pixel coordinate of black area in Contrast of the image is influenced by dynamic. segmentation image and N the pixel number of that range of gray levels in the image polarization of the. area both of them are obtained through image distribution of the black and white sharpness of edges. segmentation and period of repeating patterns It could also stands. Assuming all pixels in image as M and all image for picture quality in the narrow sence We can. pixels in dyeing area as F the proportion of calculate contrast Fcon as follows. incompletely dyed area is as follows,Fcon a4 n 10, 2 2 2 2 Tamura Texture Feature Extraction where a4 4 4 4 the fourth moments and 2.

the variance Experimentally n 1 4 was the best,Coarseness. value obtained by Tamura et al 7,Coarseness is designed to measure differences. between coarse and fine texture It provides us with Directionality. the information about the size of texture elements Directionality measures not the orientation itself but. Fine textures have smaller value of this property than the presence of it in the image It describes globally. the coarse ones We can measure coarseness of the how the texture in the image is distributed or. image by applying the following steps concentrated along certain orientations If two images. differ only in the orientation the degree of this, Step 1 Take the average at every pixel over property will be the same for them Following steps. neighborhoods whose sizes are the power of 2 are used to measure directionality. The average over the neighborhood of size 2k 2k Step 1 For each point we calculate the modulus G. k 0 1 2 3 4 5 at every pixel is and local edge direction with the formula. x 2 k 1 1 y 2 k 1 1,Ak x y i y 2k 1 g i j 2 2k G H V 2. i x 2 k 1 11,tan 1 V H 2,where g i j is the gray level at i j.

where H and V are the horizontal and vertical, Step 2 For each pixel calculate the differences elements calculated as the convolution of the image. between the not overlapping neighborhoods in with the following 3 3 operators. horizontal and vertical directions as shown below,1 0 1 1 1 1. Horizontal 1 0 1 0 0 0,Ek h x y Ak x 2 k 1 y Ak x 2 k 1 y 1 0 1 1 1 1 12. Vertical Step 2 By count all pixels with G t and, Ek v x y Ak x y 2 Ak x y 2 6 quantizing by 2k 1 2n 2k 1 2n. Step 3 For each pixel select the best size which we obtain the number of the points N k which. gives the highest output value as shown below satisfy the above constraints Then building the edge. Sbest x y 2 k 1 7,probabilities histogram HD as shown below.

where k maximizes E in either direction HD k N k i 0 N i k 0 1 n 1 13. 184 INDIAN J FIBRE TEXT RES JUNE 2016, In our experiments we used n 36 and t 12 According to the result of variance test Table 3. The histogram curve for obvious directional image the significance of rotation speed upon all features is. will exhibit a peak for images without obvious smaller than 0 05 It represents the statistics. direction it is relatively flat implication that the rotation speed has great influence. Step 3 Finally the overall direction can be obtained on all the features Besides Fcrs and Fcon the. by calculating the histogram peak sharpness This concentration has great influence on the other. measure can be defined as follows features The significance of the cross impact of. rotation speed and concentration upon H and V,Fdir p p EW p 2 H D. 14 average value and Fcon and Fdir component is smaller. than 0 05 as well which means that it has impacts on. where p is the peak value of histogram np all the them However the impact on S average value and. peak values of the histogram For each p Wp Q and Fcrs exceeds 0 05 which means it has no. represents all the bins which include it and p is the influence on them. bin which has the highest peak value,According to the result of overall test analysis. 3 Results and Discussion Tables 2 and 3 when the concentration is fixed the. 3 1 Relationship between Pattern features and Tie dye influence of rotation speed on the color is obvious. Technique The faster the rotation speeds the deeper is the color. ANOVA is a practical and valid statistics method When the rotation is fixed the influence of. which is used to calculate the significance of concentration on color is also obvious the higher the. relevant factors on the experiment results The tie dye concentration the more obvious is the color impact. experiments in this paper have two factors which can When the concentration becomes higher the impact. pose joint impact upon results To validate the of rotation speed on color becomes lower The lower. significance of interaction upon the image data the concentration the more obvious is the impact of. we made two tests for each combination of the two rotation speed on color The higher the concentration. factors the less obvious is the impact of rotation speed on. In the first test the rotation speed and the color Under the same concentration the slower the. concentration are changing and the rotation speed has rotation speed the larger is the white area Under the. three levels namely 5 circles min 15 circles min same concentration the impact of rotation speed of. and 30 circles min The concentration has five levels 30 circles min and rotation speed of 15 circles min on. Table 1 Each set of test data comprises 7 values white area and color depth is not obvious In case. including the average value of HSV tri component of of the limited change in concentration not only the. the valid tie dye area and the proportion of tie dye color impact of low concentration high rotation. white area and three Tamura features Table 2 speed but also the high concentration low rotation. Assuming that the pattern features are dependent and speed are not obvious Under different concentrations. the rotation speed and concentration are fixed with the coloring effect is better with the rotation speed of. confidence limit being 95 we obtained the ANOVA 15 circles min. Table 3 according to the multi variable analysis Table 3 shows that the former four features are. with SPSS software similar under same experiment condition but parts. of texture features have some differences such as, Table 1 Dyeing details 100g fabric No 9 10 13 15 and we can also see that their. Dye con Concentration g L images are diverse On the basis of experience it can. level be said that production process can cause this. Reactive Reactive red SSA SA,black KN B B 2BF,uncertainty rather than texture features So as.

compared to color texture features should be more,1 0 45 0 05 20 10. effective to explain pattern The Tamura coarseness. 2 0 9 0 1 30 15, is very consistent with image thickness according to. 3 2 7 0 3 40 20 visual perception that is obviously rough texture. 4 5 4 0 6 50 25 image at lower concentration its coarseness values. 5 10 8 1 2 60 30 to be significantly larger relatively very fine texture. Con Concentration the coarseness value of the image is significantly. QIONG et al TIE DYE TECHNIQUE AND PATTERN FEATURES 185. Table 2 Feature values for rotation speed and concentration dual factor repeated test variance analysis

Recent Views:

- Probabilistic risk analysis for aircraft structures with
- Navy rank abbreviations united states department of
- Universitext tracts in mathematics springer
- 5 fundamentals of health law fraud and abuse
- Interpreting newton strange beautiful
- Case ih axial flow combines cnh industrial
- Mechanical engineering in ancient egypt part xix textile
- Causal layered analysis metafuture futures studies by
- Safety and longterm accuracy of bourdon tube pressure gauges
- Compatibilite eu pro pour volkswagen