Color Image Segmentation Of Purple Soil Based On Image Feature And Adaptive Density Peaks Clustering

Purple soil riches in mineral nutrients and is the main farming soil in Southwest China.With the application of agricultural automation and sensor technology in agriculture,the requirement of applying machine vision to identify soil in the field is put forward.Purple soil color images collected in the field have complex background,including plants,weeds,scattered small soil blocks and surface soil.In order to avoid the interference of background on further processing and recognition of purple soil with machine vision,it is a primary task to segment the purple subsoil region from its background adaptively.Aiming at purple soil images taken under natural conditions,we constructed the one-dimensional separable feature.Based on the separable feature,soil region in the purple soil image was extracted by the improved density peaks clustering algorithm.The main work of this paper is as follows.(1)In order to calculate the density peaks conveniently and increase the separability between soil religion and its background,the optimization model based on maximizing the between-class variance and minimizing the within-class variance criterion was established in the Lab color space.The optimization model was solved to obtain the separable feature.Then,the improved density peaks clustering algorithm based on image histogram was utilized to cluster the gray level data of the separable feature,so as to obtain the initial segmentation result.The simulation experimental results showed that the histogram based density peaks clustering algorithm in this paper could segment the normal image samples effectively,and the initial segmentation results had higher segmentation accuracy than the comparison algorithms.(2)Aiming at the discrete small soil blocks in the background area and the internal voids in the soil area in the initial segmentation result,the post-processing algorithms of boundary extraction and region filling were designed to obtain the purple soil region image completely.The boundary extraction algorithm in this paper could extract the boundary of soil area without detecting image edge firstly.There were three times image traversal in the post-processing algorithms,which has low time complexity andtakes about 0.1s.(3)After analysis,the histogram based density peaks clustering algorithm in this paper has some problems.On the one hand,it is not robust enough to the complex purple soil image samples.On the other hand,it is unscientific and may time-consuming to determine the clustering centers manually.In the light of the shortcomings of the histogram based density peaks clustering algorithm,the global density formula was defined and a central decision measure was designed to determine the clustering centers adaptively,so that the adaptive density peaks clustering algorithm is obtained.It realized the adaptive segmentation of purple soil region,which improved the initial segmentation accuracy.20 groups normal samples set and 20 groups robust samples set were tested respectively.The simulation experimental test results show that the adaptive density peaks clustering algorithm in this paper was superior to the other 5 comparison algorithms.Its average segmentation accuracy in the initial stage was 93.45% and87.40% respectively,which was 3.16% and 12.47% higher than that of the histogram based density peaks clustering algorithm in this paper.Boundary extraction algorithm and region filling algorithm further improved the average segmentation accuracy of soil region,in the post-processing stage.In the final exact segmentation results,average segmentation accuracy was increased to 96.30% and 91.63% respectively,with average time-consuming of 0.36s and 0.35s respectively.In conclusion,the proposed algorithm is effective and robust,and it can segment purple soil images adaptively.