Color Smoke Image Segmentation Algorithm in HSI Color Space

The color smoke image segmentation algorithm based on HSI color space comprises three main stages: coarse segmentation, fine segmentation, and connected component extraction. This algorithm leverages color information in the HSI color space to partition color smoke images into distinct regions. The implementation typically involves converting RGB images to HSI color space using color transformation algorithms, where Hue and Saturation components are particularly effective for smoke detection due to their color purity and intensity independence. Initially, coarse segmentation divides the image into approximate regions using thresholding techniques on the Hue and Saturation channels. This can be implemented through histogram analysis or adaptive thresholding methods to identify potential smoke regions based on color characteristics. Subsequently, fine segmentation refines each coarse region into smaller sub-regions using more precise segmentation techniques such as region growing or watershed algorithms. This stage may incorporate edge detection operators and morphological operations to enhance boundary accuracy and remove noise. Finally, connected component extraction links adjacent sub-regions using blob analysis techniques, often implemented through algorithms like two-pass connected component labeling. This process identifies contiguous smoke regions while filtering out small isolated areas based on size and shape criteria. The algorithm effectively extracts target regions from color smoke images, providing a solid foundation for subsequent image processing and analysis tasks such as smoke density calculation and movement tracking.