Final Report Summary - COLORMATCH (Development of an expert system for the colorant formulation in the dyeing process of veneer in furniture industry)
The COLORMATCH project aimed to develop an expert system based in computer vision which will be applied in furniture factories and automatically collect and interpret data to identify and define the adequate colorant formulation for achieving a target colour. The expert system would carry a payload of state-of-the-art in furniture industry and spectrophotometers. The proposed system would be intrinsically safe (able to operate in hazardous environments) and able to carry out colorant formulations in a shorter time and with less tries, reducing considerably the needed manual intervention.
The scientific/technological objectives were to develop a computer vision system to quantify and describe the colour of non-uniformly coloured surfaces such as wood samples in furniture industry and the colour differences between them, matching with visual perception, together with algorithms that based on the data of a knowledge-base would propose a colorant formula for achieving a target colour when it is applied over a wood sample.
The result of the application of the formula should be included in the knowledge-base, thus optimising the system for achieving formulae for future use. A user interface should also form the final software, making up together with the computer vision machine an expert system for the colorant formulation for the dies applied by roller in the wood industry.
The economic and industrial objectives were:
-a significant cost reduction in the furniture industry (of the order of 704 million Euros in Europe, after project end). The development of a system to automatic colorant formulation on furniture industry should minimise losses in:
(1) formulation time (currently each formulation change involves at least 1 hour),
(2) rebuffs of wood panels (facilitating the sustainable development). Currently these losses are due to bad formulations, so this expert system should reduce colour differences between resultant sample woods and targets according to the previous quantification method.
- profits from sales of the system to reach EUR 5.5 million, four years after project end. The expected average price of this expert system will be between EUR 60 000 and 120 000.
- increase even more EU competitiveness against USA and Japanese enterprises in the furniture industry market. 2001's furniture production in Europe was EUR 77.950 million, against EUR 33.407 million of USA and EUR 18.986 million of Japan.
- besides to loss reduction other important improvements should be achieved: a final product quality improvement and the capacity to probe that quality quantitatively to the client in standard colour difference units, shorter formulation times which would reduce costs specially in the dyeing industry, as well as the capacity to formulate by any technician and not only by expert colourists that need years of experience to form. Preliminary works on this task gives us the result that actually colour differences are accepted up to about 2 or 3 E in CIELab, while our objective is to reduce these differences below to 1.5 E.
- the possibility of further investigations and developments will be take into account, and the generalisation of the method to other processes, as for example melamine printing or printing by spray gun. The method also will be applicable on other sectors like plastic industry or paper industry, which also deal with colorant formulation over non uniform surfaces and have the same problem with colour differences quantification.
The social objectives were:
- elimination of labour intensive and monotonous inspection tasks;
- reduction of the exposure of the operator to work in hazardous environments;
- elimination of operator stress and error caused by the need for great attention to detail and raw material variability;
- elimination of subjective data interpretation;
- enhancement of quality improvement in different furniture finishes;
The environmental objectives were:
- reduction of wastes of wood panels, facilitating the sustainable development. Currently, rebuffs in furniture industry born once the whole furniture piece is assembled. In case colours don't match, the whole furniture piece is rebuffed. Thus, a deviation in a furniture piece implies a similar cost of a whole assembled furniture piece, and is eliminated of the market chain and later burnt.
- reduction of the amount of employed dyes, which constitutes dangerous products, and their spillages, as well as the reduction of the exposure of furniture industry technicians to these toxic products.
The project was structured into 10 work packages (WPs), as follows:
WP1: System specification. Analysis of the current method, descriptors and factors. Specification of wood samples, varnishes, colorants and solvents;
WP2: Development of the image acquisition system. Design and construction of the prototypes and development of the image acquisition software;
WP3: Image analysis and colour coordinate transformation development. To develop the algorithms for image correction, the extraction of necessary descriptors and the transformation from camera's RGB coordinates to XYZ and Lab coordinates;
WP4: Knowledge-base preparation;
WP5: Colour differences quantification;
WP6: Development of the rules;
WP7: Development of the programme;
WP8: Final tests, validations and specifications;
WP9: Marketing study;
WP 10: Management, exploitation and dissemination.
The scientific/technological objectives were to develop a computer vision system to quantify and describe the colour of non-uniformly coloured surfaces such as wood samples in furniture industry and the colour differences between them, matching with visual perception, together with algorithms that based on the data of a knowledge-base would propose a colorant formula for achieving a target colour when it is applied over a wood sample.
The result of the application of the formula should be included in the knowledge-base, thus optimising the system for achieving formulae for future use. A user interface should also form the final software, making up together with the computer vision machine an expert system for the colorant formulation for the dies applied by roller in the wood industry.
The economic and industrial objectives were:
-a significant cost reduction in the furniture industry (of the order of 704 million Euros in Europe, after project end). The development of a system to automatic colorant formulation on furniture industry should minimise losses in:
(1) formulation time (currently each formulation change involves at least 1 hour),
(2) rebuffs of wood panels (facilitating the sustainable development). Currently these losses are due to bad formulations, so this expert system should reduce colour differences between resultant sample woods and targets according to the previous quantification method.
- profits from sales of the system to reach EUR 5.5 million, four years after project end. The expected average price of this expert system will be between EUR 60 000 and 120 000.
- increase even more EU competitiveness against USA and Japanese enterprises in the furniture industry market. 2001's furniture production in Europe was EUR 77.950 million, against EUR 33.407 million of USA and EUR 18.986 million of Japan.
- besides to loss reduction other important improvements should be achieved: a final product quality improvement and the capacity to probe that quality quantitatively to the client in standard colour difference units, shorter formulation times which would reduce costs specially in the dyeing industry, as well as the capacity to formulate by any technician and not only by expert colourists that need years of experience to form. Preliminary works on this task gives us the result that actually colour differences are accepted up to about 2 or 3 E in CIELab, while our objective is to reduce these differences below to 1.5 E.
- the possibility of further investigations and developments will be take into account, and the generalisation of the method to other processes, as for example melamine printing or printing by spray gun. The method also will be applicable on other sectors like plastic industry or paper industry, which also deal with colorant formulation over non uniform surfaces and have the same problem with colour differences quantification.
The social objectives were:
- elimination of labour intensive and monotonous inspection tasks;
- reduction of the exposure of the operator to work in hazardous environments;
- elimination of operator stress and error caused by the need for great attention to detail and raw material variability;
- elimination of subjective data interpretation;
- enhancement of quality improvement in different furniture finishes;
The environmental objectives were:
- reduction of wastes of wood panels, facilitating the sustainable development. Currently, rebuffs in furniture industry born once the whole furniture piece is assembled. In case colours don't match, the whole furniture piece is rebuffed. Thus, a deviation in a furniture piece implies a similar cost of a whole assembled furniture piece, and is eliminated of the market chain and later burnt.
- reduction of the amount of employed dyes, which constitutes dangerous products, and their spillages, as well as the reduction of the exposure of furniture industry technicians to these toxic products.
The project was structured into 10 work packages (WPs), as follows:
WP1: System specification. Analysis of the current method, descriptors and factors. Specification of wood samples, varnishes, colorants and solvents;
WP2: Development of the image acquisition system. Design and construction of the prototypes and development of the image acquisition software;
WP3: Image analysis and colour coordinate transformation development. To develop the algorithms for image correction, the extraction of necessary descriptors and the transformation from camera's RGB coordinates to XYZ and Lab coordinates;
WP4: Knowledge-base preparation;
WP5: Colour differences quantification;
WP6: Development of the rules;
WP7: Development of the programme;
WP8: Final tests, validations and specifications;
WP9: Marketing study;
WP 10: Management, exploitation and dissemination.
