Study on Color Space Conversion Based on RBF Neural Network

2010 ◽  
Vol 174 ◽  
pp. 28-31 ◽  
Author(s):  
Cong Jun Cao ◽  
Qiang Jun Liu

The conversions of color spaces are core techniques of modern ICC color management and the study of color space conversion algorithm between L*a*b* and CMYK is valuable both in theory and in application. In this paper, firstly ECI2002 standard color target data are uniformly selected, including modeling data and testing data; secondly the models of color space conversions from CMYK to L*a*b* and from L*a*b* to CMYK are built based on Radial Basis Function (RBF) neural network; finally the precision of the models are evaluated. This research indicates that the RBF neural network is suitable for the color space conversions between CMYK and L*a*b*. The models’ building processes are simpler and more convenient; the network has fast training speed and good results. With the improvement of the modeling method, this method for color space conversion will have a broader application.

2022 ◽  
pp. 004051752110672
Author(s):  
Zebin Su ◽  
Jinkai Yang ◽  
Pengfei Li ◽  
Junfeng Jing ◽  
Huanhuan Zhang

Neural networks have been widely used in color space conversion in the digital printing process. The shallow neural network easily obtains the local optimal solution when establishing multi-dimensional nonlinear mapping. In this paper, an improved high-precision deep belief network (DBN) algorithm is proposed to achieve the color space conversion from CMYK to L*a*b*. First, the PANTONE TCX color card is used as sample data, in which the CMYK value of the color block is used as input and the L*a*b* value is used as output; then, the conversion model from CMYK to L*a*b* color space is established by using DBN. To obtain better weight and threshold, DBN is optimized by a particle swarm optimization algorithm. Experimental results show that the proposed method has the highest conversion accuracy compared with Back Propagation Neural Network, Generalized Regression Neural Network, and traditional DBN color space conversion methods. It can also adapt to the actual production demand of color management in digital printing.


2010 ◽  
Vol 44-47 ◽  
pp. 3706-3710
Author(s):  
Han Kun Ye

Digital camera is the one of the main devices in the computer and multimedia technology and its color management model is the key to guarantee the color consistency in the succedent image production and transfers. The paper presents a color conversion model for digital camera based on polynomial curve generation. First, color rendering principle of digital camera is analyzed. Then digital camera data is pretreated to a unitary field to deduce final model. Third, standard color target is taken for experimental sample and substitutes color blocks in color shade district for complete color space to solve the difficulties of experimental color blocks selecting; Fourth, the model using polynomial curve generation algorithm to correct color error is deduced; Finally, the realization and experiment results show that, compared with some methods which have relatively high accuracy, the algorithm can improve color conversion accuracy and can satisfy the engineering requirement in digital camera color management


2010 ◽  
Vol 428-429 ◽  
pp. 394-397
Author(s):  
Xin Wu Li

Color management for liquid crystal display is one of the key techniques in the color image reproduction. A new color management model is presented based on overcoming flaws and limitations of current ways of liquid crystal display color management . First, the paper takes standard color target for experimental sample, and substitutes color blocks in color shade district for complete color space. Second, data collecting method is introduced and some data bases for deducing the model are created. Then, ant colony algorithm is corrected to speed up model’s convergence and a new model for liquid crystal display color management based on improved ant colony algorithm is deduced and analyzed. Finally the experimental results show that the model can improve color management accuracy of liquid crystal display and can be used in its color management practically.


2012 ◽  
Vol 461 ◽  
pp. 615-619
Author(s):  
Han Kun Ye

Color management method for photographed image files is the key to guarantee the color consistency in the succedent image production. Based on machine learning a new color management method for photographed image is advanced in the paper. First, the method selects the color shade districts of IT/8 color target as experiment sample and is taken to stand for the color space to decrease the calculation of the model. Second the method pretreats sampled data to a unitary field to satisfy the scope requirement of the input and output data of the BP neural network. Third, the active item which can adjust BP model step dynamically is used to increase convergence speed of the BP model. Finally, the experimental results indicates that the method can improve color conversion accuracy and can satisfy the engineering requirement in color management for photographed image files.


2020 ◽  
Vol 7 (3) ◽  
pp. 443
Author(s):  
Azahari Azahari ◽  
Yulindawati Yulindawati ◽  
Dewi Rosita ◽  
Syamsuddin Mallala

<p class="Abstrak">Prediksi  kelulusan  dibutuhkan  oleh  manajemen  perguruan  tinggi  dalam  menentukan kebijakan  preventif  terkait  pencegahan  dini  kasus drop  out. Lama masa studi setiap mahasiswa bisa disebabkan dengan berbagai faktor.  Dengan  menggunakan <em>data mining</em> algoritma <em>naive bayes</em> dan <em>neural network</em> dapat  dilakukan  prediksi  kelulusan  mahasiswa di  STMIK  Widya  Cipta  Dharma (WiCiDa) Samarinda . Atribut yang digunakan yaitu, umur saat masuk kuliah, klasifikasi kota asal Sekolah Menengah Atas, pekerjaan ayah, program studi, kelas, jumlah saudara, dan Indeks Prestasi Kumulatif (IPK). Sampel mahasiswa yang lulus dan <em>drop-out</em> pada tahun 2011 sampai 2019 dijadikan sebagai data <em>training</em> dan data <em>testing</em>. Sedangkan angkatan 2015–2018 digunakan sebagai data target yang akan diprediksi masa studinya. Sebanyak 3229 mahasiswa, 1769 sebagai data <em>training</em>, 321 sebagai data <em>testing</em>, dan 1139 sebagai data target. Semua data diambil dari data mahasiswa program strata 1, dan tidak mengikut sertakan data mahasiswa D3 dan alih jenjang/transfer.  Dari data <em>testing </em>diperoleh tingkat akurasi hanya 57,63%. Hasil penelitian menunjukkan banyaknya kelemahan dari hasil prediksi <em>naive bayes</em> dikarenakan tingkat akurasi kevalidannya tergolong tidak terlalu tinggi. Sedangkan akurasi prediksi <em>neural network</em> adalah 72,58%, sehingga metode alternatif inilah yang lebih baik. Proses evaluasi dan analisis dilakukan untuk melihat dimana letak kesalahan dan kebenaran dalam hasil prediksi masa studi.</p><div><div><p><em><strong>Abstract</strong></em></p><p class="Abstract"><em>Graduation predictions are required by the higher education institution preventive policies related to the early prevention of drop-out cases. The duration of study, for each student can be caused by various factors. By using the data mining algorithm Naive bayes and neural network, the student graduation in STMIK Widya Cipta Dharma (WiCiDa) can be predicted. The attributes used are as follows: age at admission, classification of cities from high school, father’s occupation, study program, class, number of siblings, and grade point average (GPA). Samples of students who graduated and dropped out between year 2011 and 2019 were used as training data and testing data. While the year class of 2015to 2018 is used as the target data, which will be predicted during the study period. According to the data mining algorithm Naive bayes, there are 3229 students; 1769 as training data, 321 as testing data, and 1139 as target data. All data is taken from students enrolled in undergraduate program and does not include data on diploma students and transfer student. From the testing data, an accuracy rate only 57.63%. The other side, prediction accuracy of the neural network is 72.58%, so this alternative method is the best chosen. The research results show the many weaknesses of the results of prediction of Naive bayes because the level of accuracy of its validity is not high. The evaluation and analysis process are conducted to see where the errors and truths are in the results of the study period predictions.</em></p><p><em><strong><br /></strong></em></p></div></div>


2019 ◽  
Vol 9 (4) ◽  
pp. 393-400
Author(s):  
Larry Pearlstein ◽  
Alexander Benasutti ◽  
Skyler Maxwell ◽  
Matthew Kilcher ◽  
Jake Bezold ◽  
...  

2012 ◽  
Vol 262 ◽  
pp. 65-68
Author(s):  
Chuan Zhi ◽  
Zhi Jian Li ◽  
Yi Shi

The nature of device color characteristic methods is the mutual conversion of device-dependent color space and device-independent color space. This paper does the comparative study on the robustness of some color space conversion methods which are based on fuzzy control, dynamic subspace divided BP neural network identification method, and fuzzy and neural identification method, by defining the robustness of color space conversion model and evaluation method. The result shows that the device color characteristic methods which are based on fuzzy and neural identification method can make the feature of BP neural network combine with fuzzy control to greatly improve the robustness of model.


Sign in / Sign up

Export Citation Format

Share Document