Satellite images have been used as a complementary information for geological studies. In order to realise the maximum potential of satillite imagery, then improvements are needed, both in the visual presentation of such images, and in their automatic classification , in order to reveal the rock differences. Methods of processing imagery, were evaluated (band ratio, principal components, decorrelated stretch and maximum likelihood) and new (canonical regression, hue-saturation-intensity HSI transform, with modified manipulation, and watershed) were evaluated with respect to their ability to reveal rock differences. It was found that the HSI method gave the best results, both for visual presentation and automatic classifcation. This method has the ability to enhance both spectral and spatial information simultaneously without any data loss which is not the case in the other image enhancement methods (band ratio, principal components or decorrelation stretch). For automatic classification, the 'hue' images produced by the HSI transformation typically gave accurate (91%) classification of all the major rock types. Further, it was shown that the watershed method of classification was superior error rate = 9% to the maximum likelihood method (error rate 14%) for the same inputs. The new method of canonical regression was evaluated and although it was not very successful, the results were encouraging and it was concluded that this method may enable the estimation of the chemical composition of exposed rocks directly from satellite imagery.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This study approaches the problem of colour image segmentation as a pattern recognition task. This leads to the problem being broken down into two component parts: feature extraction and classification algorithms. Measures to enable the objective assessment of segmentation algorithms are considered. In keeping with this pattern-recognition based philosophy, the issue of texture is approached by a consideration of features, follwed by experimentation based on classification. Techniques based on Gabor filters and fractal dimension are compared. Also colour is considered in terms of its features, and a systematic exploration of colour features in undertaken. The technique for assessing colour features is also used as the basis for a segmentation algorithm that can be used for combining colour and texture. In this study, several novel techniques are presented and discussed. Firstly a methodology for the judgement of image segmentation algorithms. Secondly a technique for segmenting images using fractal dimension is presented, including a novel application of information dimension. thirdly an objective assessment of colour spaces using the techniques discussed as the first point of this study. Finally strategies for combining colour and texture in the segmentation process are discussed and techniques presented.

Reliable colour constancy by industry for colour conducted to quantify metamerism. and metamerism indices are highly desired quality control. Two experiments were the degree of colour constancy and In the colour constancy experiment, 240 wool samples were prepared and scaled using a magnitude estimation method by a panel of 5 experienced observers under sources D65, A and TL84. 2 corresponding data sets derived from the experimental results were used to test various chromatic adaptation transforms. The results clearly show that the BFD transform gave the most precise prediction than the other transforms. Attempts were also made to derive 4 new transforms from four independent data sets. These gave similar performance as that of the BFD, but overcome the BFO's problem (incapable of predicting some of the high saturated colours). Hence, these transforms should be used with confidence for predicting the degree of colour constancy. This experimental results were also used to test various uniform colour spaces and colour appearance models. The Hunt94 model gave the most precise prediction to the colourfulness and hue results. Modification was made to its lightness scale for improving the fit. In the metamerism experiment, 76 pairs of wool samples were prepared and assessed with 20 observations using a grey scale under 7 sources: D65, A, TL84, TL83, P27, W and WW. The experimental results were used to test 3 types of illuminant metamerism indices derived here. It was found that calculating colour difference using 3 colour difference formulae, i.e. CMC, BFD and CIE94 gave the most precise prediction to the visual results. The degree of precision is quite satisfactory in comparison with typical observer precision. A new standard deviate observer (SDO) was also derived. This together with the CIE SDO and 1964 Observer were tested using the author's and the Obande's data. The results showed that the new SDO predicted results more accurate than those from the other two CIE Observers. An Observer Metamerism Index (OMI) was also derived to indicate the degree of metamerism based upon the new SDO. The results showed that the new SDO was more suitable for indicating the degree of observer metamerism.

A colour may have a different appearance under different viewing conditions. This causes many problems in the colour reproduction industry. Thus the importance of prediction of colour appearance has arisen. In this study, a mathematical model to predict colour appearance was developed based on the investigation of the changes of colour appearance under a wide range of media and viewing conditions. The media studied included large cut-sheet transparency films, 35mm projected slides, reflection samples and monitor colours. The viewing conditions varied were light source, luminance level and viewing background. Colour appearance was studied using the magnitude estimation technique. In general, colours appeared more colourful, lighter and brighter with an increase in luminance level. Background and flare light had considerable influence on colour appearance for cut-sheet transparency media. Simultaneous contrast effects occurred when a monitor colour was displayed against a chromatic surround. The monitor colour appeared lighter with a darker induction field. When a coloured area was enlarged, lightness tended to increase while colourfulness tended to decrease. Colour appearance was also affected by the closest neighbouring colour. In this case, the hue of the colour largely shifted towards the direction of the opponent hue of the induction colour. The data obtained were applied to test three colour spaces and two colour appearance models. For reflection media, the Hunt91 model performed the best. However it was not satisfactory when applied to transmissive media. Based on these results, the Hunt93 model was developed by modification of the Hunt91 model. The new model widens the application range of the Hunt91 and is reversible.